Uncertainty-aware Mean Teacher for Source-free Unsupervised Domain Adaptive Object Detection

The official pyTorch implementation of the method of Uncertainty-aware Mean Teacher for Source-free Unsupervised Domain Adaptive Object Detection

Installation

Please refer to INSTALL.md for setup.

Dataset preperation

1. Download the relevant datasets: KITTI , Waymo , nuScenes

2. Generate simulated adverse weather data for KITTI using LISA

3. Organize each folder inside data like the following

UncertaintyAwareMeanTeacher_SFUDA

├── data (main data folder)
│   ├── kitti
│   │   │── ImageSets
│   │   │── training
│   │   │   ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│   │   │── testing
│   │   │   ├──calib & velodyne & image_2
|
|
│   ├── kitti-rain
│   │   │── ImageSets
│   │   │── training
│   │   │   ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│   │   │── testing
│   │   │   ├──calib & velodyne & image_2
|
|
│   ├── kitti-snow
│   │   │── ImageSets
│   │   │── training
│   │   │   ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│   │   │── testing
│   │   │   ├──calib & velodyne & image_2
|
|
│   ├── kitti-fog
│   │   │── ImageSets
│   │   │── training
│   │   │   ├──calib & velodyne & label_2 & image_2 & (optional: planes)
│   │   │── testing
│   │   │   ├──calib & velodyne & image_2
|
|
│   ├── nuscenes
│   │   │── v1.0-trainval (or v1.0-mini if you use mini)
│   │   │   │── samples
│   │   │   │── sweeps
│   │   │   │── maps
│   │   │   │── v1.0-trainval  
|
|
│   ├── waymo
│   │   │── ImageSets
│   │   │── raw_data
│   │   │   │── segment-xxxxxxxx.tfrecord
|   |   |   |── ...
|   |   |── waymo_processed_data
│   │   │   │── segment-xxxxxxxx/
|   |   |   |── ...
│   │   │── pcdet_gt_database_train_sampled_xx/
│   │   │── pcdet_waymo_dbinfos_train_sampled_xx.pkl  
|
|

Pre-trained models

We implement the proposed method for the object detector SECOND-iou for several domain shift scenarios. You can find the folder of pretrained models here. Find specific model downloads and their corresponding config files below.

| SECOND-iou |

Go to secondiou for environment setup and implementation details.