update readme and optimize the save_dir

Author: JackWoo0831

README.md

@@ -21,6 +21,7 @@ git checkout v2  # change to v2 branch !!
 
 ## 🗺️ Latest News
 
+- ***2025.4.7*** Add more Re-ID modules (ShuffleNet, VehicleNet, MobileNet), fix some bugs (such as abandon bbox aspect ratio updating if the tracklet is not activated), and add some functions (customized low filter threshold, fuse detection score, etc.)
 - ***2025.4.3*** Support the newest ultralytics version (YOLO v3 ~ v12) and fix some bugs of hybrid sort.
 
 ## ❤️ Introduction
@@ -35,9 +36,10 @@ and the tracker supports:
 
 - SORT
 - DeepSORT 
-- ByteTrack ([ECCV2022](https://arxiv.org/pdf/2110.06864))
-- Bot-SORT ([arxiv2206](https://arxiv.org/pdf/2206.14651.pdf))
+- ByteTrack ([ECCV2022](https://arxiv.org/pdf/2110.06864)) and ByetTrack-ReID
+- Bot-SORT ([arxiv2206](https://arxiv.org/pdf/2206.14651.pdf)) and Bot-SORT-ReID
 - OCSORT ([CVPR2023](https://openaccess.thecvf.com/content/CVPR2023/papers/Cao_Observation-Centric_SORT_Rethinking_SORT_for_Robust_Multi-Object_Tracking_CVPR_2023_paper.pdf))
+- DeepOCSORT ([ICIP2023](https://arxiv.org/abs/2302.11813))
 - C_BIoU Track ([arxiv2211](https://arxiv.org/pdf/2211.14317v2.pdf))
 - Strong SORT ([IEEE TMM 2023](https://arxiv.org/pdf/2202.13514))
 - Sparse Track ([arxiv 2306](https://arxiv.org/pdf/2306.05238))
@@ -46,8 +48,17 @@ and the tracker supports:
 
 and the reid model supports:
 
+Pedestrain Re-ID:
 - OSNet
 - Extractor from DeepSort
+- ShuffleNet
+- MobileNet
+
+Vehicle Re-ID:
+- VehicleNet ([AICIty-reID-2020](https://github.com/layumi/AICIty-reID-2020))
+
+> **checkpoitns of some Re-ID models**: [Baidu Disk](https://pan.baidu.com/s/1QbVoBz4mPpf4Qsqq1PYXkQ) Code: c655
+
 
 The highlights are:
 - Supporting more trackers than MMTracking
@@ -155,11 +166,11 @@ python train_aux.py --dataset visdrone --workers 8 --device <$GPU_id$> --batch-s
 python tracker/yolo_ultralytics_utils/train_yolo_ultralytics.py --model_weight weights/yolo11m.pt --data_cfg tracker/yolo_ultralytics_utils/data_cfgs/visdrone_det.yaml --epochs 30 --batch_size 8 --img_sz 1280 --device 0
 ```
 
-
+> The training of Re-ID model please refer to its original paper or github repo. The pedestrain Re-ID model such as ShuffleNet, OSNet please refer to [torchreid](https://github.com/KaiyangZhou/deep-person-reid), the vehicle Re-ID model please refer to [AICIty-reID-2020](https://github.com/layumi/AICIty-reID-2020).
 
 ### 😊 Tracking ! 
 
-If you only want to run a demo:
+**If you only want to run a demo**:
 
 ```bash
 python tracker/track_demo.py --obj ${video path or images folder path} --detector ${yolox, yolov7 or yolo_ultra} --tracker ${tracker name} --kalman_format ${kalman format, sort, byte, ...} --detector_model_path ${detector weight path} --save_images
@@ -176,13 +187,27 @@ For example:
 python tracker/track_demo.py --obj M0203.mp4 --detector yolo_ultra_v8 --tracker deepsort --kalman_format byte --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt --save_images
 ```
 
-If you want to run trackers on dataset:
+**If you want to run trackers on dataset**:
 
 ```bash
 python tracker/track.py --dataset ${dataset name, related with the yaml file} --detector ${yolox, yolo_ultra_v8 or yolov7} --tracker ${tracker name} --kalman_format ${kalman format, sort, byte, ...} --detector_model_path ${detector weight path}
 ```
 
-For example:
+In addition, you can also specify
+
+`--reid`: Enable the reid model (currently useful for ByteTrack, BoT-SORT, OCSORT)
+
+`--reid_model`: Which model to use: Refer to `REID_MODEL_DICT` in `tracker/trackers/reid_models/engine.py` to select
+
+`--reid_model_path`: Loaded re-identification model weight path
+
+`--conf_thresh_low`: For two-stage association models (ByteTrack, BoT-SORT, etc.), the minimum confidence threshold (default 0.1)
+
+`--fuse_detection_score`: If added, the IoU value and the detection confidence value are fused, for example, the source code of BoT-SORT does this
+
+`--save_images`: Save the result image
+
+***Examples of tracking algorithms***:
 
 - SORT: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v8 --tracker sort --kalman_format sort --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt `
 

README_CN.md

@@ -15,6 +15,7 @@ git checkout v2  # change to v2 branch !!
 
 ## 🗺️ 最近更新
 
+- ***2025.4.7*** 增加更多Re-ID模型 (ShuffleNet, VehicleNet, MobileNet), 修复一些bug (例如在轨迹为非活动状态时停止更新边界框长宽), 增加一些小功能 (例如可以修改两阶段关联策略的最低阈值，原来是固定的0.1; 增加将IoU和检测置信度融合的选项)
 - ***2025.4.3*** 增加了ultralytics库最新版本的支持，修复了hybrid sort中的一些bug.
 
 
@@ -30,9 +31,10 @@ git checkout v2  # change to v2 branch !!
 
 - SORT
 - DeepSORT 
-- ByteTrack ([ECCV2022](https://arxiv.org/pdf/2110.06864))
-- Bot-SORT ([arxiv2206](https://arxiv.org/pdf/2206.14651.pdf))
+- ByteTrack ([ECCV2022](https://arxiv.org/pdf/2110.06864)) 以及 ByetTrack-ReID
+- Bot-SORT ([arxiv2206](https://arxiv.org/pdf/2206.14651.pdf)) 以及 Bot-SORT-ReID
 - OCSORT ([CVPR2023](https://openaccess.thecvf.com/content/CVPR2023/papers/Cao_Observation-Centric_SORT_Rethinking_SORT_for_Robust_Multi-Object_Tracking_CVPR_2023_paper.pdf))
+- DeepOCSORT ([ICIP2023](https://arxiv.org/abs/2302.11813))
 - C_BIoU Track ([arxiv2211](https://arxiv.org/pdf/2211.14317v2.pdf))
 - Strong SORT ([IEEE TMM 2023](https://arxiv.org/pdf/2202.13514))
 - Sparse Track ([arxiv 2306](https://arxiv.org/pdf/2306.05238))
@@ -41,8 +43,17 @@ git checkout v2  # change to v2 branch !!
 
 REID模型支持：
 
+行人重识别模型:
 - OSNet
-- DeepSORT中的
+- Extractor from DeepSort
+- ShuffleNet
+- MobileNet
+
+车辆重识别模型:
+
+- VehicleNet ([AICIty-reID-2020](https://github.com/layumi/AICIty-reID-2020))
+
+> **部分重识别模型的权重**: [百度网盘](https://pan.baidu.com/s/1QbVoBz4mPpf4Qsqq1PYXkQ) 提取码: c655
 
 亮点包括:
 - 支持的跟踪器比MMTracking多
@@ -51,10 +62,6 @@ REID模型支持：
 
 ![gif](figure/demo.gif)
 
-## 🗺️ 路线图
-
-- [ x ] Add UCMC Track
-- [] Add more ReID modules.
 
 ##  🔨 安装
 
@@ -154,11 +161,11 @@ python train_aux.py --dataset visdrone --workers 8 --device <$GPU_id$> --batch-s
 python tracker/yolo_ultralytics_utils/train_yolo_ultralytics.py --model_weight weights/yolo11m.pt --data_cfg tracker/yolo_ultralytics_utils/data_cfgs/visdrone_det.yaml --epochs 30 --batch_size 8 --img_sz 1280 --device 0
 ```
 
-
+> 关于重识别模型的训练, 请先参照对应模型的原论文或代码. 行人重识别模型例如 ShuffleNet, OSNet 参考 [torchreid](https://github.com/KaiyangZhou/deep-person-reid), 车辆重识别模型参考 [AICIty-reID-2020](https://github.com/layumi/AICIty-reID-2020).
 
 ### 😊 跟踪！
 
-如果你只是想运行一个demo:
+**如果你只是想运行一个demo**:
 
 ```bash
 python tracker/track_demo.py --obj ${video path or images folder path} --detector ${yolox, yolov7 or yolo_ultra} --tracker ${tracker name} --kalman_format ${kalman format, sort, byte, ...} --detector_model_path ${detector weight path} --save_images
@@ -175,26 +182,46 @@ python tracker/track_demo.py --obj ${video path or images folder path} --detecto
 python tracker/track_demo.py --obj M0203.mp4 --detector yolov8 --tracker deepsort --kalman_format byte --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt --save_images
 ```
 
-如果你想在数据集上测试:
+**如果你想在数据集上测试**:
 
 ```bash
 python tracker/track.py --dataset ${dataset name, related with the yaml file} --detector ${yolox, yolo_ultra_v8 or yolov7} --tracker ${tracker name} --kalman_format ${kalman format, sort, byte, ...} --detector_model_path ${detector weight path}
 ```
 
-例如:
+此外, 还可以指定
+
+`--reid`: 启用reid模型(目前对ByteTrack, BoT-SORT, OCSORT有用)
+
+`--reid_model`: 采用那种模型: 参照`tracker/trackers/reid_models/engine.py`中的`REID_MODEL_DICT`选取
+
+`--reid_model_path`: 加载的重识别模型权重路径
+
+`--conf_thresh_low`: 对于两阶段关联模型(ByteTrack, BoT-SORT等), 最低置信度阈值(默认0.1)
+
+`--fuse_detection_score`: 如果加上, 就融合IoU的值和检测置信度的值, 例如BoT-SORT的源码是这样做的
+
+`--save_images`: 保存结果图片
+
+***各种跟踪算法运行示例***:
 
 - SORT: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v8 --tracker sort --kalman_format sort --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt `
 
 - DeepSORT: `python tracker/track.py --dataset visdrone_part --detector yolov7 --tracker deepsort --kalman_format byte --detector_model_path weights/yolov8l_VisDroneDet_35epochs_20230605.pt`
 
 - ByteTrack: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v8 --tracker bytetrack --kalman_format byte --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt`
 
+- ByteTrack-ReID: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v8 --tracker bytetrack --kalman_format byte --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt --reid --reid_model osnet_x0_25 --reid_model_path weights/osnet_x0_25.pth`
+
 - OCSort: `python tracker/track.py --dataset mot17 --detector yolox --tracker ocsort --kalman_format ocsort --detector_model_path weights/bytetrack_m_mot17.pth.tar`
 
+- DeepOCSORT: `python tracker/track.py --dataset mot17 --detector yolox --tracker ocsort --kalman_format ocsort --detector_model_path weights/bytetrack_m_mot17.pth.tar --reid --reid_model shufflenet_v2_x1_0 --reid_model_path shufflenetv2_x1-5666bf0f80.pth`
+
 - C-BIoU Track: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v8 --tracker c_bioutrack --kalman_format bot --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt`
 
 - BoT-SORT: `python tracker/track.py --dataset uavdt --detector yolox --tracker botsort --kalman_format bot --detector_model_path weights/yolox_m_uavdt_50epochs.pth.tar`
 
+- BoT-SORT-ReID: `python tracker/track.py --dataset uavdt --detector yolox --tracker botsort --kalman_format bot --detector_model_path weights/yolox_m_uavdt_50epochs.pth.tar --reid --reid_model vehiclenet --reid_model_path vehicle_net_resnet50.pth`
+
 - Strong SORT: `python tracker/track.py --dataset visdrone_part --detector yolo_ultra_v8 --tracker strongsort --kalman_format strongsort --detector_model_path weights/yolov8l_VisDroneDet_35epochs_20230605.pt`
 
 - Sparse Track: `python tracker/track.py --dataset uavdt --detector yolo_ultra_v11 --tracker sparsetrack --kalman_format bot --detector_model_path weights/yolov8l_UAVDT_60epochs_20230509.pt`

tracker/track.py

@@ -115,7 +115,7 @@ def get_args():
     parser.add_argument('--gamma', type=float, default=0.1, help='param to control fusing motion and apperance dist')
     parser.add_argument('--min_area', type=float, default=150, help='use to filter small bboxs')
 
-    parser.add_argument('--save_dir', type=str, default='track_results/{dataset_name}/{split}')
+    parser.add_argument('--save_dir', type=str, default='track_results/{tracker_name}/{dataset_name}/{split}')
     parser.add_argument('--save_images', action='store_true', help='save tracking results (image)')
     parser.add_argument('--save_videos', action='store_true', help='save tracking results (video)')
 
@@ -200,7 +200,9 @@ def main(args, dataset_cfgs):
 
     logger.info(f'Total {len(seqs)} seqs will be tracked: {seqs}')
 
-    save_dir = args.save_dir.format(dataset_name=args.dataset, split=SPLIT)
+    save_dir = args.save_dir.format(tracker_name=args.tracker, dataset_name=args.dataset, split=SPLIT)
+    if not os.path.exists(save_dir):
+        os.makedirs(save_dir)
 
 
     """4. Tracking"""
@@ -292,7 +294,7 @@ def main(args, dataset_cfgs):
                 plot_img(img=ori_img, frame_id=frame_idx, results=[cur_tlwh, cur_id, cur_cls], 
                          save_dir=os.path.join(save_dir, 'vis_results'))
 
-        save_results(folder_name=os.path.join(args.dataset, SPLIT), 
+        save_results(save_dir=save_dir, 
                      seq_name=seq, 
                      results=results)
 

tracker/track_demo.py

@@ -92,6 +92,7 @@ def get_args():
     parser.add_argument('--img_size', type=int, default=1280, help='image size, [h, w]')
 
     parser.add_argument('--conf_thresh', type=float, default=0.2, help='filter tracks')
+    parser.add_argument('--conf_thresh_low', type=float, default=0.1, help='filter low conf detections, used in two-stage association')
     parser.add_argument('--nms_thresh', type=float, default=0.7, help='thresh for NMS')
     parser.add_argument('--iou_thresh', type=float, default=0.5, help='IOU thresh to filter tracks')
 

tracker/tracking_utils/tools.py

@@ -2,7 +2,7 @@
 import cv2 
 import os 
 
-def save_results(folder_name, seq_name, results, data_type='default'):
+def save_results(save_dir, seq_name, results, data_type='default'):
     """
     write results to txt file
 
@@ -12,15 +12,14 @@ def save_results(folder_name, seq_name, results, data_type='default'):
     """
     assert len(results)
 
-    if not os.path.exists(f'./track_results/{folder_name}'):
+    if not os.path.exists(save_dir):
+        os.makedirs(save_dir)
 
-        os.makedirs(f'./track_results/{folder_name}')
-
-    with open(os.path.join('./track_results', folder_name, seq_name + '.txt'), 'w') as f:
+    with open(os.path.join(save_dir, seq_name + '.txt'), 'w') as f:
         for frame_id, target_ids, tlwhs, clses, scores in results:
             for id, tlwh, score in zip(target_ids, tlwhs, scores):
                     f.write(f'{frame_id},{id},{tlwh[0]:.2f},{tlwh[1]:.2f},{tlwh[2]:.2f},{tlwh[3]:.2f},{score:.2f},-1,-1,-1\n')
 
     f.close()
 
-    return folder_name
+    return save_dir