luoduo21

同问题，我试了VOC输入都是300x300,mobilenetv1-ssd or ssdlite 都比mobilenetv2 ssdlite低2个点以上，为啥coco上v1还比v2高呢？

@dkurt Hi, do you know how to export the model with prior box node. If I add the prior box node, the error will appear: output of traced region did...

@dkurt `self.priors = self.priorbox.forward(), volatile=True)` ` output = ( self.priors, loc.view(loc.size(0), -1, 4), # loc preds self.softmax(conf.view(-1, self.num_classes)), # conf preds )` the export code: > torch.onnx.export(net, example, model_path, verbose=True,...

@dkurt I want to convert the onnx model to the ncnn model. I tried to add a Detect layer to export the post-processing directly,Although it can be successful，there are some...