weiaicunzai

> 我也有这个问题，作者的实现的差不多5个G,测试准确率有78.9， pytorch官方实现差不多1.4个G，测试集准确率也只有60%，，请问有看到问题在哪吗？ 测试代码: ``` # my implementation >>> from models.resnet import resnet50 >>> net = resnet50() >>> sum(p.numel() for p in net.parameters()) 23705252 # torchvision implementation >>> from torchvision.models...

Do you guys still have the same problem??? I've already updated my code, using Pytorch build-in module ```torch.utils.tensorboard```, and tested my code on Google Colab(python3.6, pytorch1.6), did not encounter this...

I've updated the code, you do not need to download anything now, it will automatically download cifar100 for you.

> The higher one is tested with multi-scale setting. https://github.com/hszhao/semseg/blob/master/config/cityscapes/cityscapes_pspnet50.yaml#L54 Hi, thanks for your great work, I have a question though, how did you determine the best performance weights file...

> Anchors are initial sizes (width, height) some of which (the closest to the object size) will be resized to the object size - using some outputs from the neural...

@CV-Bowen Hi, could you please tell me why the bounding boxes labels of the same cell was set to be the same? ``` boxes = tf.tile( boxes, [1, 1, 1,...

@CV-Bowen Thanks, yolov1 only predicts 1 object per cell, I know that now.

Now I can get about 86% . I think didn't download the correct mask images.

> > Hi @eraserNut, please refer to #16. The logits_mask helps to mask out self-similarity of each feature itself. Thanks for your time. I didn't quite understand why there will...

Relative Position embedding占用的显存就是比绝对位置编码多的。一个长度为5的序列，相对位置编码就有5x5个embedding，每个token要保存该token和其余所有token的相对位置编码。