jiyongma

I also encountered the same question, can I modify it to run under the tensorflow1.2 version? @TrinhQuocNguyen @yueruchen

Have you ever run into the loss = Nan problem in training VGGsegnet , or which one of your optimizers and loss functions you choose to run successfully? @vbhavank @divamgupta

When I use all the VOC data sets for training, I can run out perfectly and run the final segmentation results, but this error occurs when the number of data...

Thank you ,please give me some suggestions.

When the number of txt is reduced，/deeplab/dataset/train.txt 10582 Ｉchange it to 10580 /less

@Hakeyi 你好，我训练的时候也遇见了这个问题，训练自己的二分类数据，训练过程正常，测试图片时却全是一类结果，想问一个除了调节batch_norm_decay外，还可以怎么做，我调成0.95后还是错误 @jerryhero @PanJinquan