Connect chatgpt to this project to process user input information

[HarrisHongting]

Hi @andvg3,

I hope everything is going well with you. I'm currently working on this project and trying to put ChatGPT inside to process user input. However, I didn't find any code in the project to process user input. It seems that the whole project is just doing backend processing. I am considering creating a web app to process user input and generate scenes. Can I get some advice from you?

Kind regards,

Harris

[andvg3]

Hi,

I think it is an amazing idea to inject ChatGPT response to our LSDM model. You can do it by simply replace the input text with the ChatGPT prompts, with some decent engineering work.

Nevertheless, I have to suggest you that it would be very challenging because LSDM is an offline model and it took a few minutes to generate a scene, which may be not useful in such interactive case.

Please let me know if you have any additional questions. Thank you!

[HarrisHongting]

Hi @andvg3,

python fit_custom_obj.py
--sequence_name <sequence_name>
--vertices_path data/supp/proxd_valid/vertices/<sequence_name>_verts.npy
--contact_labels_path data/supp/proxd_valid/semantics/<sequence_name>_cfs.npy
--output_dir fitting_results/
--label <num_label>
--file_name training/sdm/output/predictions/<interaction_name>.npy

In this instruction, I found that the related .npy files of <sequence_name>_verts.npy and <sequence_name>_cfs.npy in the related path are all composed of array lists. And their names are some incomprehensible numbers. This is very confusing to me. Could you please take some time to explain it to me?

Many thanks,

Harris

[andvg3]

Hi, the .npy files are array of 3D points, which is, actually, the point cloud of the human motion.