search

lufficc / SSD

High quality, fast, modular reference implementation of SSD in PyTorch
MIT License
1.52k stars 384 forks source link

How to visualization? #163

Closed Keeping7 closed 4 years ago

Keeping7 commented 4 years ago

@lufficc Thank you for your good work. I want to know how to show the APs,APm,APl, mAP for COCO and every categories' AP for VOC dataset. Namely, how to use tensorboard in pytorch to show realtime results? Thank you.

lufficc commented 4 years ago

Currently this repo has already implemented these. See: https://github.com/lufficc/SSD/blob/master/ssd/data/datasets/evaluation/coco/__init__.py#L48-L53 https://github.com/lufficc/SSD/blob/master/ssd/data/datasets/evaluation/voc/__init__.py#L44-L50 You can run tensorboard --logdir <outputs-dir> --port 6006 and open http://127.0.0.1:6006 to check the visualization.

Keeping7 commented 4 years ago

@lufficc Yeah. Will Tensorboard opened in training phase or test phase or both? When i run tensorboard --logdir='./outputs/vgg_ssd300_voc0712/tf_logs' --port 6006 , meanwhile a problem has arisen. The pictures are as follows, image image How to solve this problem? Thank you.

lufficc commented 4 years ago

Tensorboard is opened only in training phase. I don't know why you have this problem but you can try the solutions below:

  1. use --logdir="./outputs/vgg_ssd300_voc0712"(Double quotes) instead of --logdir='./outputs/vgg_ssd300_voc0712/tf_logs'
  2. upgrade tensorboard.
Keeping7 commented 4 years ago

@lufficc Yeah, it works. Thank you for your help.

bellzhong677 commented 3 years ago

Hello, I don't know why I don't have this folder tf_logs after training