Skip to content

Files

Latest commit

Lan-pingLan-ping
add newfusion patch & fusion patch
Jun 12, 2022
053dacd · Jun 12, 2022

History

History

transfuser

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
assets
assets
May 25, 2022
README.md
README.md
May 25, 2022
config.py
config.py
Jun 12, 2022
data.py
data.py
May 25, 2022
data_viz.py
data_viz.py
May 25, 2022
model.py
model.py
Jun 4, 2022
model_viz.py
model_viz.py
May 25, 2022
requirements.txt
requirements.txt
May 25, 2022
train.py
train.py
May 25, 2022
train.sh
train.sh
May 25, 2022
viz.py
viz.py
May 25, 2022

README.md

TransFuser

TransFuser uses the self-attention mechanism of the transformers for image and LiDAR feature fusion at multiple resolutions.

Training

CUDA_VISIBLE_DEVICES=<gpu_id> python3 train.py --id transfuser --batch_size 56

Evaluation

Update leaderboard/scripts/run_evaluation.sh to include the following.

export ROUTES=leaderboard/data/evaluation_routes/routes_town05_long.xml
export TEAM_AGENT=leaderboard/team_code/transfuser_agent.py
export TEAM_CONFIG=model_ckpt/transfuser
export CHECKPOINT_ENDPOINT=results/transfuser_result.json
export SCENARIOS=leaderboard/data/scenarios/town05_all_scenarios.json