enricobu96 / mymid

Implementation of Stochastic Trajectory Prediction via Motion Indeterminacy Diffusion - Gu et al. Improvement of model performances by implementing Denoising Diffusion Probabilistic Models (Ho et al.)

Official Implementation for CVPR2022 Paper "Stochastic Trajectory Prediction via Motion Indeterminacy Diffusion"

By Tianpei Gu*, Guangyi Chen*, Junlong Li, Chunze Lin, Yongming Rao, Jie Zhou and Jiwen Lu

Environment

We use pytorch 1.7.1 with cuda > 10.1 for all experiments.

Prepare Data

python process_data.py

Train & Test

First modify or create your own config file in /configs and run python main.py --config configs/YOUR_CONFIG.yaml --dataset DATASET where $DATASET should from ["eth", "hotel", "univ", "zara1", "zara2", "sdd"]

Documentation

In order to get documentation for the project, install the following pip modules:

pip install handsdown mkdocs

Then generate the .md documentation files with:

python -m handsdown --external `git config --get remote.origin.url` --create-configs

And then build and serve the documentation with:

python -m mkdocs build
python -m mkdocs serve

The documentation is now available on http://127.0.0.1/8000/.

mymid's Issues

Change average of two nets into linear layer

Thesis

• Re-read methods
• Write experiments and results
• Write conclusion
• Check consistency between equations
• Re-read everything
  • Improving vocabulary
  • Adding acronyms to the list
  • Fix italic, bold, texttt
• Fix references
• Write dedication, acknowledgments
• Fix images

Implement the goal also for the other losses