who who who

What is the value of the mAP on VOC dataset?

You should try it and show me the mAP

Ok, thanks hejinIzhanghao 于2019年9月5日周四 下午4:12写道： OK. Once my machines have spare time. And the model is exactly same as the one in r1.13. And the weights file trained by r1.13...

1. Migrate to tf2 2. Set num_epochs=1 and num_examples_per_epoch=num_epochs * num_exampels

Please specific eval_samples too. eval_batch_size=64

You should change anchor size.

Try to change https://github.com/google/automl/blob/2dbfb7984c20c24d856b96b54498799ed1b270cb/efficientdet/aug/autoaugment.py#L316 to ```python return image_ops.rotate(mask, angle, interpolation=self.interpolation, fill_mode='reflect') ```

Training longer and specificing ```--val_json_file```. Don't forget to change ```aspect_ratios``` if necessary https://github.com/google/automl/issues/412.

> Use standard file utilities to get mtimes. > ERROR:tensorflow:Model diverged with loss = NaN. > E0513 16:02:25.483811 139845663950592 basic_session_run_hooks.py:760] Model diverged with loss = NaN. > ERROR:tensorflow:Error recorded from...

Reference tf1 debug tutorial.