ECE 285 trajectory prediction project

Instructions

Go to the Nuscenes github page and install the devkit. Once the devkit is installed, download the data set from the Nuscenes website. In our experiments, we were only able to download the first 85 scenes from the TrainVal dataset.
Use the semantic segmentation network here to segment the camera images in the TrainVal dataset. Ensure that the destination directory for the semantically segmented camera images is different than the source directory.
When segmentation is finished, rename the source directory where the camera originated to a different name such as CAM_FRONT_OLD and rename the target directory to the previous source destination name. This step is important because it allows us to access the segmented camera images using the functions provided by the nuscenes devkit.
Construct the intermediate representation using generate_data.py.
Train the model using train_safety.py

You can read about our findings here!

Name		Name	Last commit message	Last commit date
Latest commit History 30 Commits
models		models
trained_model_weights		trained_model_weights
.DS_Store		.DS_Store
BaselineDataset.py		BaselineDataset.py
ConvLSTM.py		ConvLSTM.py
KittiDataset.py		KittiDataset.py
LayerNorm2D.py		LayerNorm2D.py
Model.py		Model.py
Nuscenes_dataset.py		Nuscenes_dataset.py
Nuscenes_dataset_test.py		Nuscenes_dataset_test.py
README.md		README.md
args.py		args.py
baseline-transfer.ipynb		baseline-transfer.ipynb
baseline.ipynb		baseline.ipynb
data_preprocessing_final.ipynb		data_preprocessing_final.ipynb
data_preprocessing_map.ipynb		data_preprocessing_map.ipynb
generate_data.py		generate_data.py
infer-main.ipynb		infer-main.ipynb
infer-transfer.ipynb		infer-transfer.ipynb
logs		logs
nuscenes_main.ipynb		nuscenes_main.ipynb
oxford-test.ipynb		oxford-test.ipynb
prediction_tutorial.ipynb		prediction_tutorial.ipynb
train-safety.py		train-safety.py
train.py		train.py