Evaluation on the synthetic data with copenet_singleview

[Samuel3Shin]

Hi! I'm trying to run the evaluation on the synthetic data part.

cd AirPose/copenet_real
python src/copenet_real/scripts/hmr_synth_res_compile.py "model type" "precalculated results directory Path" "/path to the dataset" "your_path/AirPose/copenet/src/copenet/data/smplx/models/smplx"

I downloaded all the correct pre-trained models and datasets.
But I get some errors in the rendering part when I run the above code because of hardcoded absolute paths
(https://github.com/robot-perception-group/AirPose/blob/main/copenet_real/src/copenet_real/scripts/hmr_synth_res_compile.py#L118-L119)

I think you used some other preprocessed dataset to render the results. How can we preprocess data to render it with your rendering module?

My goal is to render the results of my trained single-view model with real data. Therefore, I'm trying to do the above task!

Thanks!

[nitin-ppnp]

Could you please try by replacing the path string with datapath?

[Samuel3Shin]

Thank you so much that worked!

Now, I'm trying to get rendering results from copenet_dji_real_data with the copenet_singleview model.
I'm trying to tweak the script file "/AirPose/copenet_real/src/copenet_real/scripts/hmr_synth_res_compile.py" to handle "copenet_dji_real_data". But currently, I met a lot of errors.

Do you think it is a good approach??

Or do you think tweaking "/AirPose/copenet_real_data/scripts/bundle_adj.py" script to use the copenet_singleview model is better??

Thanks, Nitin!