# parameters nc: 80 # number of classes depth_multiple: 1.0 # expand model depth width_multiple: 1.0 # expand layer channels # anchors anchors: - [13,17, 31,25, 24,51, 61,45] # P3/8 - [61,45, 48,102, 119,96, 97,189] # P4/16 - [97,189, 217,184, 171,384, 324,451] # P5/32 - [324,451, 545,357, 616,618, 1024,1024] # P6/64 # csp-p6 backbone backbone: # [from, number, module, args] [[-1, 1, Conv, [32, 3, 1]], # 0 [-1, 1, Conv, [64, 3, 2]], # 1-P1/2 [-1, 1, BottleneckCSP, [64]], [-1, 1, Conv, [128, 3, 2]], # 3-P2/4 [-1, 3, BottleneckCSP, [128]], [-1, 1, Conv, [256, 3, 2]], # 5-P3/8 [-1, 15, BottleneckCSP, [256]], [-1, 1, Conv, [512, 3, 2]], # 7-P4/16 [-1, 15, BottleneckCSP, [512]], [-1, 1, Conv, [1024, 3, 2]], # 9-P5/32 [-1, 7, BottleneckCSP, [1024]], [-1, 1, Conv, [1024, 3, 2]], # 11-P6/64 [-1, 7, BottleneckCSP, [1024]], # 12 ] # yolov4-p6 head # na = len(anchors[0]) head: [[-1, 1, SPPCSP, [512]], # 13 [-1, 1, Conv, [512, 1, 1]], [-1, 1, nn.Upsample, [None, 2, 'nearest']], [-6, 1, Conv, [512, 1, 1]], # route backbone P5 [[-1, -2], 1, Concat, [1]], [-1, 3, BottleneckCSP2, [512]], # 18 [-1, 1, Conv, [256, 1, 1]], [-1, 1, nn.Upsample, [None, 2, 'nearest']], [-13, 1, Conv, [256, 1, 1]], # route backbone P4 [[-1, -2], 1, Concat, [1]], [-1, 3, BottleneckCSP2, [256]], # 23 [-1, 1, Conv, [128, 1, 1]], [-1, 1, nn.Upsample, [None, 2, 'nearest']], [-20, 1, Conv, [128, 1, 1]], # route backbone P3 [[-1, -2], 1, Concat, [1]], [-1, 3, BottleneckCSP2, [128]], # 28 [-1, 1, Conv, [256, 3, 1]], [-2, 1, Conv, [256, 3, 2]], [[-1, 23], 1, Concat, [1]], # cat [-1, 3, BottleneckCSP2, [256]], # 32 [-1, 1, Conv, [512, 3, 1]], [-2, 1, Conv, [512, 3, 2]], [[-1, 18], 1, Concat, [1]], # cat [-1, 3, BottleneckCSP2, [512]], # 36 [-1, 1, Conv, [1024, 3, 1]], [-2, 1, Conv, [512, 3, 2]], [[-1, 13], 1, Concat, [1]], # cat [-1, 3, BottleneckCSP2, [512]], # 40 [-1, 1, Conv, [1024, 3, 1]], [[29,33,37,41], 1, Detect, [nc, anchors]], # Detect(P3, P4, P5, P6) ]