Added GIoU

build/darknet/x64/cfg/yolov3-voc.yolov3-giou-40.cfg

@@ -0,0 +1,808 @@
+[net]
+# Testing
+# batch=1
+# subdivisions=1
+# Training
+batch=64
+subdivisions=16
+width=416
+height=416
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+## single gpu
+learning_rate=0.001
+burn_in=1000
+max_batches = 100400
+
+## 2x
+#learning_rate=0.0005
+#burn_in=2000
+#max_batches = 100400
+#max_batches = 200800
+
+## 4x
+#learning_rate=0.00025
+#burn_in=4000
+#max_batches = 50200
+##max_batches = 200800
+
+policy=steps
+steps=40000,45000
+scales=.1,.1
+
+
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+######################
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 6,7,8
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 61
+
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 3,4,5
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 36
+
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 0,1,2
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+

build/darknet/x64/cfg/yolov3.coco-giou-12.cfg

@@ -0,0 +1,806 @@
+[net]
+# Testing
+# batch=1
+# subdivisions=1
+# Training
+batch=64
+subdivisions=16
+width=608
+height=608
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+## single gpu
+learning_rate=0.001
+burn_in=1000
+max_batches = 550400
+
+## 2 gpu
+#learning_rate=0.0005
+#burn_in=2000
+#max_batches = 500200
+
+## 4 gpu
+#learning_rate=0.00025
+#burn_in=4000
+#max_batches = 500200
+###max_batches = 2000800
+
+policy=steps
+steps=400000,450000
+scales=.1,.1
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+######################
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 6,7,8
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 61
+
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 3,4,5
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou
+
+
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 36
+
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 0,1,2
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou

cfg/yolov3-voc.yolov3-giou-40.cfg

@@ -0,0 +1,808 @@
+[net]
+# Testing
+# batch=1
+# subdivisions=1
+# Training
+batch=64
+subdivisions=16
+width=416
+height=416
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+## single gpu
+learning_rate=0.001
+burn_in=1000
+max_batches = 100400
+
+## 2x
+#learning_rate=0.0005
+#burn_in=2000
+#max_batches = 100400
+#max_batches = 200800
+
+## 4x
+#learning_rate=0.00025
+#burn_in=4000
+#max_batches = 50200
+##max_batches = 200800
+
+policy=steps
+steps=40000,45000
+scales=.1,.1
+
+
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+######################
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 6,7,8
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 61
+
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 3,4,5
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 36
+
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=75
+activation=linear
+
+[yolo]
+mask = 0,1,2
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=20
+num=9
+jitter=.3
+ignore_thresh = .5
+truth_thresh = 1
+random=1
+iou_normalizer=0.25
+cls_normalizer=1.0
+iou_loss=giou
+

cfg/yolov3.coco-giou-12.cfg

@@ -0,0 +1,806 @@
+[net]
+# Testing
+# batch=1
+# subdivisions=1
+# Training
+batch=64
+subdivisions=16
+width=608
+height=608
+channels=3
+momentum=0.9
+decay=0.0005
+angle=0
+saturation = 1.5
+exposure = 1.5
+hue=.1
+
+## single gpu
+learning_rate=0.001
+burn_in=1000
+max_batches = 550400
+
+## 2 gpu
+#learning_rate=0.0005
+#burn_in=2000
+#max_batches = 500200
+
+## 4 gpu
+#learning_rate=0.00025
+#burn_in=4000
+#max_batches = 500200
+###max_batches = 2000800
+
+policy=steps
+steps=400000,450000
+scales=.1,.1
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=3
+stride=1
+pad=1
+activation=leaky
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=32
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=64
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+# Downsample
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=2
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=1024
+size=3
+stride=1
+pad=1
+activation=leaky
+
+[shortcut]
+from=-3
+activation=linear
+
+######################
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=512
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=1024
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 6,7,8
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 61
+
+
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=256
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=512
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 3,4,5
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou
+
+
+
+[route]
+layers = -4
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[upsample]
+stride=2
+
+[route]
+layers = -1, 36
+
+
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+filters=128
+size=1
+stride=1
+pad=1
+activation=leaky
+
+[convolutional]
+batch_normalize=1
+size=3
+stride=1
+pad=1
+filters=256
+activation=leaky
+
+[convolutional]
+size=1
+stride=1
+pad=1
+filters=255
+activation=linear
+
+
+[yolo]
+mask = 0,1,2
+anchors = 10,13,  16,30,  33,23,  30,61,  62,45,  59,119,  116,90,  156,198,  373,326
+classes=80
+num=9
+jitter=.3
+ignore_thresh = .7
+truth_thresh = 1
+random=1
+iou_normalizer=0.5
+iou_loss=giou

include/darknet.h

@@ -105,6 +105,11 @@ typedef enum {
     LOGISTIC, RELU, RELIE, LINEAR, RAMP, TANH, PLSE, LEAKY, ELU, LOGGY, STAIR, HARDTAN, LHTAN, SELU
 }ACTIVATION;
 
+// parser.h
+typedef enum {
+    IOU, GIOU, MSE
+} IOU_LOSS;
+
 // image.h
 typedef enum{
     PNG, BMP, TGA, JPG
@@ -309,6 +314,10 @@ struct layer {
     float *weights;
     float *weight_updates;
 
+    float iou_normalizer;
+    float cls_normalizer;
+    int iou_loss;
+
     char *align_bit_weights_gpu;
     float *mean_arr_gpu;
     float *align_workspace_gpu;
@@ -671,6 +680,24 @@ typedef struct box {
     float x, y, w, h;
 } box;
 
+// box.h
+typedef struct boxabs {
+    float left, right, top, bot;
+} boxabs;
+
+// box.h
+typedef struct dxrep {
+    float dt, db, dl, dr;
+} dxrep;
+
+// box.h
+typedef struct ious {
+    float iou, giou;
+    dxrep dx_iou;
+    dxrep dx_giou;
+} ious;
+
+
 // box.h
 typedef struct detection{
     box bbox;

src/box.c

@@ -64,6 +64,30 @@ dbox derivative(box a, box b)
     return d;
 }
 
+// where c is the smallest box that fully encompases a and b
+boxabs box_c(box a, box b) {
+    boxabs ba = { 0 };
+    ba.top = fmin(a.y - a.h / 2, b.y - b.h / 2);
+    ba.bot = fmax(a.y + a.h / 2, b.y + b.h / 2);
+    ba.left = fmin(a.x - a.w / 2, b.x - b.w / 2);
+    ba.right = fmax(a.x + a.w / 2, b.x + b.w / 2);
+    return ba;
+}
+
+// representation from x, y, w, h to top, left, bottom, right
+boxabs to_tblr(box a) {
+    boxabs tblr = { 0 };
+    float t = a.y - (a.h / 2);
+    float b = a.y + (a.h / 2);
+    float l = a.x - (a.w / 2);
+    float r = a.x + (a.w / 2);
+    tblr.top = t;
+    tblr.bot = b;
+    tblr.left = l;
+    tblr.right = r;
+    return tblr;
+}
+
 float overlap(float x1, float w1, float x2, float w2)
 {
     float l1 = x1 - w1/2;
@@ -93,7 +117,113 @@ float box_union(box a, box b)
 
 float box_iou(box a, box b)
 {
-    return box_intersection(a, b)/box_union(a, b);
+    //return box_intersection(a, b)/box_union(a, b);
+
+    float I = box_intersection(a, b);
+    float U = box_union(a, b);
+    if (I == 0 || U == 0) {
+        return 0;
+    }
+    return I / U;
+}
+
+float box_giou(box a, box b)
+{
+    boxabs ba = box_c(a, b);
+    float w = ba.right - ba.left;
+    float h = ba.bot - ba.top;
+    float c = w*h;
+    float iou = box_iou(a, b);
+    if (c == 0) {
+        return iou;
+    }
+    float u = box_union(a, b);
+    float giou_term = (c - u) / c;
+#ifdef DEBUG_PRINTS
+    printf("  c: %f, u: %f, giou_term: %f\n", c, u, giou_term);
+#endif
+    return iou - giou_term;
+}
+
+dxrep dx_box_iou(box pred, box truth, IOU_LOSS iou_loss) {
+    boxabs pred_tblr = to_tblr(pred);
+    float pred_t = fmin(pred_tblr.top, pred_tblr.bot);
+    float pred_b = fmax(pred_tblr.top, pred_tblr.bot);
+    float pred_l = fmin(pred_tblr.left, pred_tblr.right);
+    float pred_r = fmax(pred_tblr.left, pred_tblr.right);
+
+    boxabs truth_tblr = to_tblr(truth);
+#ifdef DEBUG_PRINTS
+    printf("\niou: %f, giou: %f\n", box_iou(pred, truth), box_giou(pred, truth));
+    printf("pred: x,y,w,h: (%f, %f, %f, %f) -> t,b,l,r: (%f, %f, %f, %f)\n", pred.x, pred.y, pred.w, pred.h, pred_tblr.top, pred_tblr.bot, pred_tblr.left, pred_tblr.right);
+    printf("truth: x,y,w,h: (%f, %f, %f, %f) -> t,b,l,r: (%f, %f, %f, %f)\n", truth.x, truth.y, truth.w, truth.h, truth_tblr.top, truth_tblr.bot, truth_tblr.left, truth_tblr.right);
+#endif
+    //printf("pred (t,b,l,r): (%f, %f, %f, %f)\n", pred_t, pred_b, pred_l, pred_r);
+    //printf("trut (t,b,l,r): (%f, %f, %f, %f)\n", truth_tblr.top, truth_tblr.bot, truth_tblr.left, truth_tblr.right);
+    dxrep dx = { 0 };
+    float X = (pred_b - pred_t) * (pred_r - pred_l);
+    float Xhat = (truth_tblr.bot - truth_tblr.top) * (truth_tblr.right - truth_tblr.left);
+    float Ih = fmin(pred_b, truth_tblr.bot) - fmax(pred_t, truth_tblr.top);
+    float Iw = fmin(pred_r, truth_tblr.right) - fmax(pred_l, truth_tblr.left);
+    float I = Iw * Ih;
+    float U = X + Xhat - I;
+
+    float Cw = fmax(pred_r, truth_tblr.right) - fmin(pred_l, truth_tblr.left);
+    float Ch = fmax(pred_b, truth_tblr.bot) - fmin(pred_t, truth_tblr.top);
+    float C = Cw * Ch;
+
+    // float IoU = I / U;
+    // Partial Derivatives, derivatives
+    float dX_wrt_t = -1 * (pred_r - pred_l);
+    float dX_wrt_b = pred_r - pred_l;
+    float dX_wrt_l = -1 * (pred_b - pred_t);
+    float dX_wrt_r = pred_b - pred_t;
+
+    // gradient of I min/max in IoU calc (prediction)
+    float dI_wrt_t = pred_t > truth_tblr.top ? (-1 * Iw) : 0;
+    float dI_wrt_b = pred_b < truth_tblr.bot ? Iw : 0;
+    float dI_wrt_l = pred_l > truth_tblr.left ? (-1 * Ih) : 0;
+    float dI_wrt_r = pred_r < truth_tblr.right ? Ih : 0;
+    // derivative of U with regard to x
+    float dU_wrt_t = dX_wrt_t - dI_wrt_t;
+    float dU_wrt_b = dX_wrt_b - dI_wrt_b;
+    float dU_wrt_l = dX_wrt_l - dI_wrt_l;
+    float dU_wrt_r = dX_wrt_r - dI_wrt_r;
+    // gradient of C min/max in IoU calc (prediction)
+    float dC_wrt_t = pred_t < truth_tblr.top ? (-1 * Cw) : 0;
+    float dC_wrt_b = pred_b > truth_tblr.bot ? Cw : 0;
+    float dC_wrt_l = pred_l < truth_tblr.left ? (-1 * Ch) : 0;
+    float dC_wrt_r = pred_r > truth_tblr.right ? Ch : 0;
+
+    // Final IOU loss (prediction) (negative of IOU gradient, we want the negative loss)
+    float p_dt = 0;
+    float p_db = 0;
+    float p_dl = 0;
+    float p_dr = 0;
+    if (U > 0) {
+        p_dt = ((U * dI_wrt_t) - (I * dU_wrt_t)) / (U * U);
+        p_db = ((U * dI_wrt_b) - (I * dU_wrt_b)) / (U * U);
+        p_dl = ((U * dI_wrt_l) - (I * dU_wrt_l)) / (U * U);
+        p_dr = ((U * dI_wrt_r) - (I * dU_wrt_r)) / (U * U);
+    }
+
+    if (iou_loss == GIOU) {
+        if (C > 0) {
+            // apply "C" term from gIOU
+            p_dt += ((C * dU_wrt_t) - (U * dC_wrt_t)) / (C * C);
+            p_db += ((C * dU_wrt_b) - (U * dC_wrt_b)) / (C * C);
+            p_dl += ((C * dU_wrt_l) - (U * dC_wrt_l)) / (C * C);
+            p_dr += ((C * dU_wrt_r) - (U * dC_wrt_r)) / (C * C);
+        }
+    }
+
+    // apply grad from prediction min/max for correct corner selection
+    dx.dt = pred_tblr.top < pred_tblr.bot ? p_dt : p_db;
+    dx.db = pred_tblr.top < pred_tblr.bot ? p_db : p_dt;
+    dx.dl = pred_tblr.left < pred_tblr.right ? p_dl : p_dr;
+    dx.dr = pred_tblr.left < pred_tblr.right ? p_dr : p_dl;
+
+    return dx;
 }
 
 float box_rmse(box a, box b)

src/box.h

@@ -33,7 +33,10 @@ extern "C" {
 box float_to_box(float *f);
 float box_iou(box a, box b);
 float box_rmse(box a, box b);
+dxrep dx_box_iou(box a, box b, IOU_LOSS iou_loss);
+float box_giou(box a, box b);
 dbox diou(box a, box b);
+boxabs to_tblr(box a);
 void do_nms(box *boxes, float **probs, int total, int classes, float thresh);
 void do_nms_sort_v2(box *boxes, float **probs, int total, int classes, float thresh);
 //LIB_API void do_nms_sort(detection *dets, int total, int classes, float thresh);

src/option_list.c

@@ -112,6 +112,13 @@ char *option_find_str(list *l, char *key, char *def)
     return def;
 }
 
+char *option_find_str_quiet(list *l, char *key, char *def)
+{
+    char *v = option_find(l, key);
+    if (v) return v;
+    return def;
+}
+
 int option_find_int(list *l, char *key, int def)
 {
     char *v = option_find(l, key);

src/option_list.h

@@ -18,6 +18,7 @@ int read_option(char *s, list *options);
 void option_insert(list *l, char *key, char *val);
 char *option_find(list *l, char *key);
 char *option_find_str(list *l, char *key, char *def);
+char *option_find_str_quiet(list *l, char *key, char *def);
 int option_find_int(list *l, char *key, int def);
 int option_find_int_quiet(list *l, char *key, int def);
 float option_find_float(list *l, char *key, float def);

src/parser.c

@@ -333,7 +333,15 @@ layer parse_yolo(list *options, size_params params)
     }
     //assert(l.outputs == params.inputs);
 
-    //l.max_boxes = option_find_int_quiet(options, "max", 90);
+    l.iou_normalizer = option_find_float_quiet(options, "iou_normalizer", 0.75);
+    l.cls_normalizer = option_find_float_quiet(options, "cls_normalizer", 1);
+    char *iou_loss = option_find_str_quiet(options, "iou_loss", "mse");   //  "iou");
+
+    if (strcmp(iou_loss, "mse") == 0) l.iou_loss = MSE;
+    else if (strcmp(iou_loss, "giou") == 0) l.iou_loss = GIOU;
+    else l.iou_loss = IOU;
+    fprintf(stderr, "Yolo layer params: iou loss: %s, iou_normalizer: %f, cls_normalizer: %f\n", (l.iou_loss == MSE ? "mse" : (l.iou_loss == GIOU ? "giou" : "iou")), l.iou_normalizer, l.cls_normalizer);
+
     l.jitter = option_find_float(options, "jitter", .2);
     l.focal_loss = option_find_int_quiet(options, "focal_loss", 0);
 

src/utils.c

@@ -621,6 +621,19 @@ float mag_array(float *a, int n)
     return sqrt(sum);
 }
 
+// indicies to skip is a bit array
+float mag_array_skip(float *a, int n, int * indices_to_skip)
+{
+    int i;
+    float sum = 0;
+    for (i = 0; i < n; ++i) {
+        if (indices_to_skip[i] != 1) {
+            sum += a[i] * a[i];
+        }
+    }
+    return sqrt(sum);
+}
+
 void scale_array(float *a, int n, float s)
 {
     int i;

src/utils.h

@@ -58,6 +58,7 @@ float mean_array(float *a, int n);
 void mean_arrays(float **a, int n, int els, float *avg);
 float variance_array(float *a, int n);
 float mag_array(float *a, int n);
+float mag_array_skip(float *a, int n, int * indices_to_skip);
 float dist_array(float *a, float *b, int n, int sub);
 float **one_hot_encode(float *a, int n, int k);
 float sec(clock_t clocks);

src/yolo_layer.c

@@ -127,7 +127,7 @@ box get_yolo_box(float *x, float *biases, int n, int index, int i, int j, int lw
     b.h = exp(x[index + 3*stride]) * biases[2*n+1] / h;
     return b;
 }
-
+/*
 float delta_yolo_box(box truth, float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, float *delta, float scale, int stride)
 {
     box pred = get_yolo_box(x, biases, n, index, i, j, lw, lh, w, h, stride);
@@ -144,7 +144,57 @@ float delta_yolo_box(box truth, float *x, float *biases, int n, int index, int i
     delta[index + 3*stride] = scale * (th - x[index + 3*stride]);
     return iou;
 }
+*/
 
+ious delta_yolo_box(box truth, float *x, float *biases, int n, int index, int i, int j, int lw, int lh, int w, int h, float *delta, float scale, int stride, float iou_normalizer, IOU_LOSS iou_loss)
+{
+    ious all_ious = { 0 };
+    // i - step in layer width
+    // j - step in layer height
+    //  Returns a box in absolute coordinates
+    box pred = get_yolo_box(x, biases, n, index, i, j, lw, lh, w, h, stride);
+    all_ious.iou = box_iou(pred, truth);
+    all_ious.giou = box_giou(pred, truth);
+    // avoid nan in dx_box_iou
+    if (pred.w == 0) { pred.w = 1.0; }
+    if (pred.h == 0) { pred.h = 1.0; }
+    if (iou_loss == MSE)    // old loss
+    {
+        float tx = (truth.x*lw - i);
+        float ty = (truth.y*lh - j);
+        float tw = log(truth.w*w / biases[2 * n]);
+        float th = log(truth.h*h / biases[2 * n + 1]);
+
+        delta[index + 0 * stride] = scale * (tx - x[index + 0 * stride]);
+        delta[index + 1 * stride] = scale * (ty - x[index + 1 * stride]);
+        delta[index + 2 * stride] = scale * (tw - x[index + 2 * stride]);
+        delta[index + 3 * stride] = scale * (th - x[index + 3 * stride]);
+    }
+    else {
+        // https://github.com/generalized-iou/g-darknet
+        // https://arxiv.org/abs/1902.09630v2
+        // https://giou.stanford.edu/
+        all_ious.dx_iou = dx_box_iou(pred, truth, iou_loss);
+
+        // jacobian^t (transpose)
+        delta[index + 0 * stride] = (all_ious.dx_iou.dl + all_ious.dx_iou.dr);
+        delta[index + 1 * stride] = (all_ious.dx_iou.dt + all_ious.dx_iou.db);
+        delta[index + 2 * stride] = ((-0.5 * all_ious.dx_iou.dl) + (0.5 * all_ious.dx_iou.dr));
+        delta[index + 3 * stride] = ((-0.5 * all_ious.dx_iou.dt) + (0.5 * all_ious.dx_iou.db));
+
+        // predict exponential, apply gradient of e^delta_t ONLY for w,h
+        delta[index + 2 * stride] *= exp(x[index + 2 * stride]);
+        delta[index + 3 * stride] *= exp(x[index + 3 * stride]);
+
+        // normalize iou weight
+        delta[index + 0 * stride] *= iou_normalizer;
+        delta[index + 1 * stride] *= iou_normalizer;
+        delta[index + 2 * stride] *= iou_normalizer;
+        delta[index + 3 * stride] *= iou_normalizer;
+    }
+
+    return all_ious;
+}
 
 void delta_yolo_class(float *output, float *delta, int index, int class_id, int classes, int stride, float *avg_cat, int focal_loss)
 {
@@ -218,7 +268,11 @@ void forward_yolo_layer(const layer l, network_state state)
 
     memset(l.delta, 0, l.outputs * l.batch * sizeof(float));
     if (!state.train) return;
-    float avg_iou = 0;
+    //float avg_iou = 0;
+    float tot_iou = 0;
+    float tot_giou = 0;
+    float tot_iou_loss = 0;
+    float tot_giou_loss = 0;
     float recall = 0;
     float recall75 = 0;
     float avg_cat = 0;
@@ -253,19 +307,19 @@ void forward_yolo_layer(const layer l, network_state state)
                     }
                     int obj_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 4);
                     avg_anyobj += l.output[obj_index];
-                    l.delta[obj_index] = 0 - l.output[obj_index];
+                    l.delta[obj_index] = l.cls_normalizer * (0 - l.output[obj_index]);
                     if (best_iou > l.ignore_thresh) {
                         l.delta[obj_index] = 0;
                     }
                     if (best_iou > l.truth_thresh) {
-                        l.delta[obj_index] = 1 - l.output[obj_index];
+                        l.delta[obj_index] = l.cls_normalizer * (1 - l.output[obj_index]);
 
                         int class_id = state.truth[best_t*(4 + 1) + b*l.truths + 4];
                         if (l.map) class_id = l.map[class_id];
                         int class_index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 4 + 1);
                         delta_yolo_class(l.output, l.delta, class_index, class_id, l.classes, l.w*l.h, 0, l.focal_loss);
                         box truth = float_to_box_stride(state.truth + best_t*(4 + 1) + b*l.truths, 1);
-                        delta_yolo_box(truth, l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, state.net.w, state.net.h, l.delta, (2-truth.w*truth.h), l.w*l.h);
+                        delta_yolo_box(truth, l.output, l.biases, l.mask[n], box_index, i, j, l.w, l.h, state.net.w, state.net.h, l.delta, (2 - truth.w*truth.h), l.w*l.h, l.iou_normalizer, l.iou_loss);
                     }
                 }
             }
@@ -299,11 +353,19 @@ void forward_yolo_layer(const layer l, network_state state)
             int mask_n = int_index(l.mask, best_n, l.n);
             if (mask_n >= 0) {
                 int box_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 0);
-                float iou = delta_yolo_box(truth, l.output, l.biases, best_n, box_index, i, j, l.w, l.h, state.net.w, state.net.h, l.delta, (2-truth.w*truth.h), l.w*l.h);
+                //float iou =
+                ious all_ious = delta_yolo_box(truth, l.output, l.biases, best_n, box_index, i, j, l.w, l.h, state.net.w, state.net.h, l.delta, (2 - truth.w*truth.h), l.w*l.h, l.iou_normalizer, l.iou_loss);
+
+                // range is 0 <= 1
+                tot_iou += all_ious.iou;
+                tot_iou_loss += 1 - all_ious.iou;
+                // range is -1 <= giou <= 1
+                tot_giou += all_ious.giou;
+                tot_giou_loss += 1 - all_ious.giou;
 
                 int obj_index = entry_index(l, b, mask_n*l.w*l.h + j*l.w + i, 4);
                 avg_obj += l.output[obj_index];
-                l.delta[obj_index] = 1 - l.output[obj_index];
+                l.delta[obj_index] = l.cls_normalizer * (1 - l.output[obj_index]);
 
                 int class_id = state.truth[t*(4 + 1) + b*l.truths + 4];
                 if (l.map) class_id = l.map[class_id];
@@ -312,14 +374,54 @@ void forward_yolo_layer(const layer l, network_state state)
 
                 ++count;
                 ++class_count;
-                if(iou > .5) recall += 1;
-                if(iou > .75) recall75 += 1;
-                avg_iou += iou;
+                //if(iou > .5) recall += 1;
+                //if(iou > .75) recall75 += 1;
+                //avg_iou += iou;
+                if (all_ious.iou > .5) recall += 1;
+                if (all_ious.iou > .75) recall75 += 1;
             }
         }
     }
+    //*(l.cost) = pow(mag_array(l.delta, l.outputs * l.batch), 2);
+    //printf("Region %d Avg IOU: %f, Class: %f, Obj: %f, No Obj: %f, .5R: %f, .75R: %f,  count: %d\n", state.index, avg_iou / count, avg_cat / class_count, avg_obj / count, avg_anyobj / (l.w*l.h*l.n*l.batch), recall / count, recall75 / count, count);
+
+    float avg_iou_loss = 0;
+    // gIOU loss + MSE (objectness) loss
+    if (l.iou_loss == MSE) {
         *(l.cost) = pow(mag_array(l.delta, l.outputs * l.batch), 2);
-    printf("Region %d Avg IOU: %f, Class: %f, Obj: %f, No Obj: %f, .5R: %f, .75R: %f,  count: %d\n", state.index, avg_iou/count, avg_cat/class_count, avg_obj/count, avg_anyobj/(l.w*l.h*l.n*l.batch), recall/count, recall75/count, count);
+    }
+    else {
+        // Always compute classification loss both for iou + cls loss and for logging with mse loss
+        // TODO: remove IOU loss fields before computing MSE on class
+        //   probably split into two arrays
+        int stride = l.w*l.h;
+        float* no_iou_loss_delta = calloc(l.batch * l.outputs, sizeof(float));
+        memcpy(no_iou_loss_delta, l.delta, l.batch * l.outputs * sizeof(float));
+        for (b = 0; b < l.batch; ++b) {
+            for (j = 0; j < l.h; ++j) {
+                for (i = 0; i < l.w; ++i) {
+                    for (n = 0; n < l.n; ++n) {
+                        int index = entry_index(l, b, n*l.w*l.h + j*l.w + i, 0);
+                        no_iou_loss_delta[index + 0 * stride] = 0;
+                        no_iou_loss_delta[index + 1 * stride] = 0;
+                        no_iou_loss_delta[index + 2 * stride] = 0;
+                        no_iou_loss_delta[index + 3 * stride] = 0;
+                    }
+                }
+            }
+        }
+        float classification_loss = l.cls_normalizer * pow(mag_array(no_iou_loss_delta, l.outputs * l.batch), 2);
+        free(no_iou_loss_delta);
+
+        if (l.iou_loss == GIOU) {
+            avg_iou_loss = count > 0 ? l.iou_normalizer * (tot_giou_loss / count) : 0;
+        }
+        else {
+            avg_iou_loss = count > 0 ? l.iou_normalizer * (tot_iou_loss / count) : 0;
+        }
+        *(l.cost) = avg_iou_loss + classification_loss;
+    }
+    printf("v3 (%s loss, Normalizer: (iou: %f, cls: %f) Region %d Avg (IOU: %f, GIOU: %f), Class: %f, Obj: %f, No Obj: %f, .5R: %f, .75R: %f, count: %d\n", (l.iou_loss == MSE ? "mse" : (l.iou_loss == GIOU ? "giou" : "iou")), l.iou_normalizer, l.cls_normalizer, state.index, tot_iou / count, tot_giou / count, avg_cat / class_count, avg_obj / count, avg_anyobj / (l.w*l.h*l.n*l.batch), recall / count, recall75 / count, count);
 }
 
 void backward_yolo_layer(const layer l, network_state state)