Datasets

We provide a data index in CSV format for all the data used from the MS2 dataset, including the training, validation, and test sets. Please download the dataset and place the data into the MS2_dataset directory.

Trained Models

Download the trained models for baseline approaches, including early, intermediate, and late fusion, and our method via the link

Running

• Test
  

Download a trained model that you want to test and put it into ./runs/. Then, use the following scripts to evaluate.

Methods	Description
test_EF.py	evaluating the early fusion of vision, LiDAR, and thermal data
test_IF.py	evaluating the intermediate fusion of vision, LiDAR, and thermal data
test_LF.py	evaluating the late fusion of vision, LiDAR, and thermal data
test_our.py	evaluating the proposed method for fusion of vision, LiDAR, and thermal data
test_all_ta.py	evaluating the proposed method for fusion of vision, LiDAR, and thermal data with the introduced trustworthiness assessment method

• Train

Use the following scripts to train different methods.

Methods	Description
train_EF.py	training the early fusion model of vision, LiDAR, and thermal data
train_IF.py	training the intermediate fusion model of vision, LiDAR, and thermal data
train_LF.py	training the late fusion model of vision, LiDAR, and thermal data
train_v.py	training the model for only vision
train_l.py	training the model for only lidar
train_t.py	training the model for only thermal
train_vl.py	training the model for vision-lidar fusion
train_lt.py	training the model for lidar-thermal fusion
train_vt.py	training the model for vision-thermal fusion
train_all_our.py	training the proposed model for fusion of vision, LiDAR, and thermal data