search

Song-Jingyu / PointPainting

This repository is an open-source PointPainting package which is easy to understand, deploy and run!
MIT License
243 stars 46 forks source link

could you please tell me how to run the test set? #3

Closed 1553545365 closed 3 years ago

1553545365 commented 3 years ago

Could you please tell me how to run the test set?

I run : "python test.py --cfg_file cfgs/kitti_models/pointpillar_painted.yaml --batch_size 1 --ckpt /home/wenhao/桌面/PointPainting-main/detector/output/kitti_models/pointpillar_painted/default/ckpt/checkpoint_epoch_80.pth".

But the result of "pointpillar+DeepLab V3+" seemed have no different with original "pointpillar"

Song-Jingyu commented 3 years ago

Hi,

There should be at least some difference with original PointPillar. When you use the PointPillar, you may verify if you have used the original ckpt trained by PointPillar config.

1553545365 commented 3 years ago

Hi,

There should be at least some difference with original PointPillar. When you use the PointPillar, you may verify if you have used the original ckpt trained by PointPillar config.

wowowow! Thank you for your reply. Do you mean my following command correct? ( "python test.py --cfg_file cfgs/kitti_models/pointpillar_painted.yaml --batch_size 1 --ckpt /home/wenhao/桌面/PointPainting-main/detector/output/kitti_models/pointpillar_painted/default/ckpt/checkpoint_epoch_80.pth")

Song-Jingyu commented 3 years ago

I think it should be right! Actually I did not run test.py. The test.py is actually doing evaluation, which is similar to the output of training. I believe your command is correct.

1553545365 commented 3 years ago

I think it should be right! Actually I did not run test.py. The test.py is actually doing evaluation, which is similar to the output of training. I believe your command is correct.

In fact, I've got the results of the test set, but I'm not sure if the results have been Pointpainted? This is my test datapath: detector ├── data │ ├── kitti │ │ │── ImageSets │ │ │── training │ │ │ ├── calib │ │ │ ├── image_2 │ │ │ ├── image_3 │ │ │ ├── label_2 │ │ │ ├── velodyne │ │ │ ├── planes │ │ │ ├── painted_lidar │ │ │── testing │ │ │ ├── calib │ │ │ ├── image_2 │ │ │ ├── velodyne

Do I need to run painting.py for the test set, before I running test.py? Thank you soooo mach.

Song-Jingyu commented 3 years ago

Hi,

I am quite busy these days. I think you might misunderstand the test.py function, it still aims to use the validation set to test, not test set. I did not check how to test capability on test set. I think you may check OpenPCDet for more details.

Best, Jingyu