ludas

首先感谢无私分享，我有个疑问：在你的upsample_layer.cpp的使用的是双线性插值，正常的双线性插值对齐应该写成onst Dtype b1_x = (x+0.5)/w_ratio_ - 0.5，而你的65行写成onst Dtype b1_x = (x+1)/w_ratio_ - 1.0;使用的1.0会导致对齐偏向图像右下角，也就是特征图的左上角值不会参与使用，这样做的目的是什么，还是我理解有问题？

your model doesn't match to your deploy.txt

4

when run"./build/examples/cpp_recognition/recognition.bin data/captcha/1.png examples/crnn/deploy.prototxt examples/crnn/model/crnn_captcha.caffemodel",error show "Cannot copy param 0 weights from layer 'conv0'; shape mismatch. Source param shape is 64 3 3 3 (1728); target param shape is 64...

Hello, I would like to consult. Now I have tensorRT reasoning model of fp16, how to use fp16 infer of onnx-tensorRT