maohule

iteration 0, test infinitely

1

I use the TVG_CRFRNN.sh which at caffe/examples/segmentationcrfasrnn/ to train crfasrnn.But the training iterate 0 and then test infinitely as follows. What should I do?Thanks! I0102 21:24:32.579792 5267 upgrade_proto.cpp:620] Attempting to...

You initialised your network on CPU, please initialize it on GPU.

1

Hello @bittnt I train crfasrnn with GPU using the latest caffe as [https://github.com/bittnt/caffe/tree/crfrnn](url). I exactly follow the instruction of [https://github.com/martinkersner/train-CRF-RNN](url).But when I run python solve.py 2>&1 | tee train.log,there have...

iteration 0, test infinitely

3

Hi @martinkersner I use the TVG_CRFRNN.sh which at caffe/examples/segmentationcrfasrnn/ to train crfasrnn.But the training iterate 0 and then test infinitely as follows. What should I do?Thanks! I0102 21:24:32.579792 5267 upgrade_proto.cpp:620]...

Does it support Opencv3 and Cuda8.0?

您好，感谢您分享的论文代码。这里有个问题想要询问一下。dataset.py中target['labels'] = torch.ones([1, num_points]).squeeze(0).type(torch.LongTensor)，这里明确把人类别标签设为1。但是在conditional_detr.py文件中，loss_cardinality函数中的代码card_pred = (pred_logits.argmax(-1) != pred_logits.shape[-1] - 1).sum(1)（第169行）。通过计算，pred_logits.shape[-1] - 1等于1。那么不应该是pred_logits.argmax(-1) == pred_logits.shape[-1] - 1).sum(1)为预测的人数吗？