YOLOv5 Series Multi-backbone(TPH-YOLOv5, Ghostnet, ShuffleNetv2, Mobilenetv3Small, EfficientNetLite, PP-LCNet, SwinTransformer YOLO), Module(CBAM, DCN), Pruning (EagleEye, Network Slimming), Quantization (MQBench) and Deployment (TensorRT, ncnn) Compression Tool Box.
994
stars
200
forks
source link
大佬，如果我想把模型改小一点，改成yolov5sP2CBAM-Swin-BiFPN-SPP.yaml的话，除了调整width和depth的乘参数外还有别的调整方式嘛？有没有可能存在更适用于小网络的改法嘞？