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.
998
stars
202
forks
source link
关于YOLOv5x-TPH的结果和模型大小 #70
Open
like121 opened 2 years ago
您好,请问YOLOv5x-TPH的结果也是在val验证集上得到的吗?还有就是对应YOLOv5x-TPH的yaml文件是哪个呢?我尝试了一些yaml文件都没有得到对应大小的模型,自己复现的TPH-YOLOv5x则是GFLOPS达到了九百多....