lambdayin
lambdayin
https://github.com/ruinmessi/ASFF/blob/master/models/yolov3_head.py(line 134). "return refined_pred.data" => "return refined_pred" (when you convert pth to onnx)
> When doing QAT with xnn.quantize in edgeai-torchvision, weights are discretized (internally) and compensated for the accuracy loss due to that discretization. Although, the weights saved in the checkpoint are...
源代码没有问题呀，T1w表示的是从世界坐标系到萝卜1号坐标系的转换。 T2w * T1w.inverse() * p1 就是先T1w.inverse() * p1=Tw1*p1把该点从1号坐标系转到世界坐标系，T2w * (T1w.inverse() * p1) 然后再从世界坐标系转到2号坐标系。
我和你结果差不多，对极约束算下来也大部分在10-3量级，个别误匹配的比较大