search

OpenGVLab / PonderV2

PonderV2: Pave the Way for 3D Foundation Model with A Universal Pre-training Paradigm
https://arxiv.org/abs/2310.08586
MIT License
324 stars 6 forks source link

Question about experiment folder created by training script. #16

Closed GhaitsaDina closed 5 months ago

GhaitsaDina commented 6 months ago

Hello, thank you for your great work! I have collection of point cloud data that I want to test using your pre-trained model checkpoint. Can I use them without training the S3DIS dataset?

I'm a bit confused because from reading the testing script, there's an experiment folder that I couldn't fine anywhere except making it by training S3DIS myself. How can I test my own data without training the whole S3DIS dataset first? image

Your guidance is very much needed, thank you so much.

HaoyiZhu commented 6 months ago

Hi, thanks for your interest! To test on you own dataset, you can directly refer to tools/test.py, which is just called by scripts/test.sh (see the last few lines):

$PYTHON -u tools/$TEST_CODE \
  --config-file "$CONFIG_DIR" \
  --num-gpus "$GPU" \
  --options save_path="$EXP_DIR" weight="${MODEL_DIR}"/"${WEIGHT}".pth

Also, you should write your own config, dataset loader and tester of your task. You can refer to our examples such as ScanNet test config, ScanNet test data loading and semantic segmentation tester.

As you can see in our SemSegTester, the basic logic is to 1) build and load model 2) build dataloader 3) loop over dataloader and get model predictions 4) compute metrics and post-processing

HaoyiZhu commented 6 months ago

Further instructions can be found here