### Install SimpleTrack Library

Source: https://github.com/tusen-ai/simpletrack/blob/main/README.md

Installs the SimpleTrack algorithm as a local library named `mot_3d`. This command assumes you are in the root directory of the project.

```bash
pip install -e ./
```

--------------------------------

### Run SimpleTrack Demo

Source: https://github.com/tusen-ai/simpletrack/blob/main/README.md

Executes the demo for SimpleTrack using preprocessed data. It requires specifying the demo name, detector name, object type, configuration path, and data folder. The `--visualize` flag enables interactive visualization.

```bash
python tools/demo.py \
    --name demo \
    --det_name cp \
    --obj_type vehicle \
    --config_path configs/waymo_configs/vc_kf_giou.yaml \
    --data_folder ./demo_data/ \
    --visualize
```

--------------------------------

### Preprocess Waymo Raw Data

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Converts Waymo Open Dataset tf_records into a more usable format. Requires raw data folder, target data folder, and an optional number of processes for faster execution.

```bash
cd preprocessing/waymo_data
bash waymo_preprocess.sh ${raw_data_folder} ${data_folder} ${process_num}
```

--------------------------------

### Preprocess nuScenes Raw Data

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Extracts data from nuScenes, supporting both key frames (2Hz) and all frames (20Hz). Requires the raw data directory and target directories for 2Hz and 20Hz data.

```bash
cd preprocessing/nuscenes_data
bash nuscenes_preprocess.sh ${raw_data_dir} ${data_dir_2hz} ${data_dir_20hz}
```

--------------------------------

### Decode Waymo Ground Truth

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Decodes Waymo Open Dataset ground truth .bin files into a format suitable for 3D MOT and Detection. Requires the path to the .bin file and the target data folder.

```python
cd preprocessing/waymo_data
python gt_bin_decode.py --data_folder ${data_dir} --file_path ${bin_path}
```

--------------------------------

### Preprocess nuScenes Detection Results (20Hz)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Converts nuScenes 20Hz detection files (JSON) into .npz format. Requires raw data folder, 20Hz data folder, detection name, detection file path, and the 'velo' flag if velocity data is needed.

```python
cd preprocessing/nuscenes_data
python detection.py --raw_data_folder ${raw_data_dir} --data_folder ${data_dir_20hz} --det_name ${name} --file_path ${file_path} --mode 20hz --velo
```

--------------------------------

### Preprocess nuScenes Detection Results (2Hz)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Converts nuScenes 2Hz detection files (JSON) into .npz format. Requires raw data folder, 2Hz data folder, detection name, detection file path, and the 'velo' flag if velocity data is needed.

```python
cd preprocessing/nuscenes_data
python detection.py --raw_data_folder ${raw_data_dir} --data_folder ${data_dir_2hz} --det_name ${name} --file_path ${file_path} --mode 2hz --velo
```

--------------------------------

### Run SimpleTrack Inference on Waymo Dataset (Cyclist)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/waymo.md

Executes the SimpleTrack inference for cyclists on the Waymo Open Dataset. It uses a specific configuration file for cyclist tracking and requires detection names, data and result folders, and the number of processes for multiprocessing. The results are saved in the specified result folder.

```bash
python tools/main_waymo.py \
    --name SimpleTrack \
    --det_name ${det_name} \
    --obj_type vehicle \
    --config_path configs/waymo_configs/vc_kf_giou.yaml \
    --data_folder ${waymo_data_dir} \
    --result_folder ${waymo_result_dir} \
    --process ${proc_num}
```

--------------------------------

### SimpleTrack Configuration Parameters (YAML)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/config.md

This YAML configuration defines various parameters for the SimpleTrack algorithm, including association settings, motion models, and data loading options. It controls aspects like score thresholds, association metrics (IoU, GIoU), motion model type, and Non-Maximum Suppression (NMS).

```yaml
running:
  covariance: default         # not used
  score_threshold: 0.7        # detection score threshold for first-stage association and output
  max_age_since_update: 2     # count for death, same as AB3DMOT
  min_hits_to_birth: 3        # count for birth, same as AB3DMOT
  match_type: bipartite       # use hungarian (biparitite) or greedy algorithm (greedy) for association
  asso: giou                  # association metric, we support GIoU (giou) and IoU (iou) for now
  has_velo: false
  motion_model: kf            # Kalman filter (kf) as motion model
  asso_thres:
    iou: 0.9                  # association threshold, 1 - IoU has to be smaller than it.
    giou: 1.5                 # association threshold, 1 - GIoU has to be smaller than it.

redundancy:
  mode: mm                    # (mm) denotes two-stage association, (default) denotes one stage association (see nuScenes configs)
  det_score_threshold:
    iou: 0.1
    giou: 0.1
  det_dist_threshold:
    iou: 0.1                  # IoU has to be greater than this
    giou: -0.5                # GIoU has to be greater than this

data_loader:
  pc: true                    # load point clouds for visualization
  nms: true                   # apply NMS for data preprocessing, True or False
  nms_thres: 0.25             # IoU-3D threshold for NMS
```

--------------------------------

### Convert SimpleTrack Results to Waymo Open Dataset .bin Format

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/waymo.md

Converts the tracking results generated by SimpleTrack into the official Waymo Open Dataset .bin format. This is necessary for using official evaluation tools. It requires the name of the tracking approach, the result folder, and the data folder.

```bash
python tools/waymo_pred_bin.py \
    --name SimpleTrack \
    --result_folder ${waymo_result_dir} \
    --data_folder ${waymo_data_dir}
```

--------------------------------

### Run SimpleTrack Inference on Waymo Dataset (Pedestrian)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/waymo.md

Executes the SimpleTrack inference for pedestrians on the Waymo Open Dataset. This command is similar to the vehicle inference but uses a different configuration file for pedestrian tracking. It requires detection names, configuration paths, data and result folders, and the number of processes for multiprocessing.

```bash
python tools/main_waymo.py \
    --name SimpleTrack \
    --det_name ${det_name} \
    --obj_type vehicle \
    --config_path configs/waymo_configs/pd_kf_giou.yaml \
    --data_folder ${waymo_data_dir} \
    --result_folder ${waymo_result_dir} \
    --process ${proc_num}
```

--------------------------------

### Preprocess Waymo Detection Results

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/data_preprocess.md

Preprocesses Waymo detection results for 3D MOT inference. Takes detection file path, target data folder, and an optional name for the detection. Metadata and ID flags can be included.

```python
cd preprocessing/waymo_data
python detection.py --name ${name} --data_folder ${data_dir} --file_path ${bin_path} --metadata --id
```

--------------------------------

### Run SimpleTrack Inference on Waymo Dataset (Vehicle)

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/waymo.md

Executes the SimpleTrack inference for vehicles on the Waymo Open Dataset. It requires detection names, configuration paths, data and result folders, and the number of processes for multiprocessing. The results are saved in the specified result folder.

```bash
python tools/main_waymo.py \
    --name SimpleTrack \
    --det_name ${det_name} \
    --obj_type vehicle \
    --config_path configs/waymo_configs/vc_kf_giou.yaml \
    --data_folder ${waymo_data_dir} \
    --result_folder ${waymo_result_dir} \
    --gt_folder ${gt_dets_dir} \
    --process ${proc_num}
```

--------------------------------

### Python Citation for SimpleTrack

Source: https://github.com/tusen-ai/simpletrack/blob/main/README.md

Provides the BibTeX entry for citing the SimpleTrack paper in academic publications. This is useful for acknowledging the research in your own work.

```bibtex
@article{pang2021simpletrack,
  title={SimpleTrack: Understanding and Rethinking 3D Multi-object Tracking},
  author={Pang, Ziqi and Li, Zhichao and Wang, Naiyan},
  journal={arXiv preprint arXiv:2111.09621},
  year={2021}
}
```

--------------------------------

### Run SimpleTrack 10Hz Inference

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/nuScenes.md

Executes SimpleTrack inference for the 10Hz setting on the nuScenes dataset, as proposed in the paper. Requires similar parameters as the 2Hz setting, with results saved in a dedicated folder.

```bash
python tools/main_nuscenes_10hz.py \
    --name SimpleTrack10Hz \
    --det_name ${det_name} \
    --config_path configs/nu_configs/giou.yaml \
    --result_folder ${nuscenes_result_dir} \
    --data_folder ${nuscenes20hz_data_dir} \
    --process ${proc_num}
```

--------------------------------

### Run SimpleTrack 2Hz Inference

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/nuScenes.md

Executes SimpleTrack inference for the 2Hz (key-frame) setting on the nuScenes dataset. Requires specifying detection name, config path, result folder, data folder, and processing number. Results are saved per sequence in the specified result folder.

```bash
python tools/main_nuscenes.py \
    --name SimpleTrack2Hz \
    --det_name ${det_name} \
    --config_path configs/nu_configs/giou.yaml \
    --result_folder ${nuscenes_result_dir} \
    --data_folder ${nuscenes2hz_data_dir} \
    --process ${proc_num}
```

--------------------------------

### Convert SimpleTrack 2Hz Results to nuScenes JSON

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/nuScenes.md

Converts the inference results from SimpleTrack's 2Hz setting into the official nuScenes `.json` format. This involves two steps: creating the results and then merging by type.

```bash
python tools/nuscenes_result_creation.py \
    --name SimpleTrack2Hz \
    --result_folder ${nuscenes_result_dir} \
    --data_folder ${nuscenes2hz_data_dir}
```

```bash
python tools/nuscenes_type_merge.py \
    --name SimpleTrack2Hz \
    --result_folder ${nuscenes_result_dir}
```

--------------------------------

### Convert SimpleTrack 10Hz Results to nuScenes JSON

Source: https://github.com/tusen-ai/simpletrack/blob/main/docs/nuScenes.md

Converts the inference results from SimpleTrack's 10Hz setting into the official nuScenes `.json` format. This process is similar to the 2Hz conversion, using a different script for result creation.

```bash
python tools/nuscenes_result_creation_10hz.py \
    --name SimpleTrack10Hz \
    --result_folder ${nuscenes_result_dir} \
    --data_folder ${nuscenes20hz_data_dir}
```

```bash
python tools/nuscenes_type_merge.py \
    --name SimpleTrack10Hz \
    --result_folder ${nuscenes_result_dir}
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.