Fix AttributeError: 'numpy.float32' object has no attribute 'int'

[romanab]

In version NumPy 1.24.0, the deprecated np.int has been removed.

[mikel-brostrom]

The hard requirement on numpy comes from trackeval: JonathonLuiten/TrackEval#117. Unless it is updated this package cannot be pushed

[romanab]

Ok. At least it’s here for whom ever needs to change it in their own install. Your whole repo works so far for me except for that.

…

On Mon, Jun 12, 2023 at 2:36 AM Mike ***@***.***> wrote: The hard requirement on numpy comes from trackeval: JonathonLuiten/TrackEval#117 <JonathonLuiten/TrackEval#117>. Unless it is updated this package cannot be pushed — Reply to this email directly, view it on GitHub <#937 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABH75HQSHCYHN4RHFLV5OBTXK3PLVANCNFSM6AAAAAAZC465UQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[mikel-brostrom]

This could be because you are using python<3.8?

[romanab]

Don't think so. The first time around, I downloaded the repo & tried running it under an Anaconda environment of Python 3.9.16. At some point I saw a message about some component just being supported under linux so instead of figuring it out, I took the easy way out. I downloaded a new Ubuntu install to run under WSL2 and installed again an environment under Anaconda of Python 3.10.11. Both times I was getting the following traceback with strongsort (everything else was fine; though at some point there was a mention that the examples/weights directory was missing which I just then created): Traceback (most recent call last): File "/home/roman/yolo_tracking/examples/track.py", line 222, in <module> main(opt) File "/home/roman/yolo_tracking/examples/track.py", line 217, in main run(vars(opt)) File "/home/roman/anaconda3/envs/yolo-tracking/lib/python3.10/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context return func(*args, **kwargs) File "/home/roman/yolo_tracking/examples/track.py", line 126, in run predictor.tracker_outputs[i] = predictor.trackers[i].update(dets.cpu().detach().numpy(), im0) File "/home/roman/yolo_tracking/boxmot/strongsort/strong_sort.py", line 64, in update self.tracker.update(detections, clss, confs) File "/home/roman/yolo_tracking/boxmot/strongsort/sort/tracker.py", line 95, in update self.tracks[track_idx].update( File "/home/roman/yolo_tracking/boxmot/strongsort/sort/track.py", line 279, in update self.class_id = class_id.int() AttributeError: 'numpy.float32' object has no attribute 'int' Applying the previous fix allowed me to run strongsort component which previously wouldn't run.

…

On Tue, Jun 13, 2023 at 1:31 AM Mike ***@***.***> wrote: This could be because you are using python<3.8? — Reply to this email directly, view it on GitHub <#937 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABH75HVTWTISH3F2R5UTDVLXLAQPTANCNFSM6AAAAAAZC465UQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[meua]

This could be because you are using python<3.8?

python 3.8 has the same problem，as follows:

(py38) jarvis@jia:~/coding/pyhome/mikel-brostrom/yolo_tracking$ python examples/evolve.py --tracking-method strongsort --benchmark MOT17 --n-trials 100 
evolve: yolo_model=/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/weights/yolov8n.pt, reid_model=/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/weights/lmbn_n_cuhk03_d.pt, tracking_method=strongsort, name=exp, project=/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/runs/evolve, classes=['0'], exist_ok=False, benchmark=MOT17, split=train, eval_existing=, conf=0.45, imgsz=[1280], device=[''], n_trials=100, resume=False, processes_per_device=2, objectives=['HOTA', 'MOTA', 'IDF1'], tracking_config=/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/boxmot/strongsort/configs/strongsort.yaml
[I 2023-06-15 16:56:30,725] A new study created in memory with name: no-name-9d0dc866-21aa-4152-a09a-d13379e1b243
2023-06-15 16:56:30.727 | INFO     | __main__:get_new_config:185 - Writing newly generated config for trial
2023-06-15 16:56:30.739 | INFO     | val:download_mot_eval_tools:64 - Eval repo already downloaded
2023-06-15 16:56:30.740 | INFO     | val:eval:195 - Staring evaluation process on /home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/val_utils/data/MOT17/train/MOT17-02-FRCNN/MOT17-02-FRCNN
2023-06-15 16:56:35.030 | ERROR    | val:eval:223 - 2023-06-15 16:56:32.418 | INFO     | boxmot.utils.torch_utils:select_device:52 - Yolo Tracking v10.0.13 🚀 Python-3.8.16 torch-1.13.1+cu117 CUDA:0 (NVIDIA GeForce RTX 3080 Ti, 12037MiB)

2023-06-15 16:56:32.418 | INFO     | __main__:run:66 - {'yolo_model': PosixPath('/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/weights/yolov8n.pt'), 'reid_model': PosixPath('/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/weights/lmbn_n_cuhk03_d.pt'), 'tracking_method': 'strongsort', 'source': '/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/val_utils/data/MOT17/train/MOT17-02-FRCNN/MOT17-02-FRCNN', 'imgsz': [1280], 'conf': 0.4706543672687518, 'iou': 0.7, 'device': '', 'show': False, 'save': True, 'classes': [0], 'project': '/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/runs/evolve', 'name': 'exp2', 'exist_ok': True, 'half': False, 'vid_stride': 1, 'hide_label': False, 'hide_conf': False, 'save_txt': True}
2023-06-15 16:56:34.563 | INFO     | __main__:run:171 - image 1/600 /home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/val_utils/data/MOT17/train/MOT17-02-FRCNN/MOT17-02-FRCNN/000001.jpg: 736x1280 12 persons, YOLO 7.7ms, TRACKING 24.6ms
Traceback (most recent call last):
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/track.py", line 222, in <module>
    main(opt)
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/track.py", line 217, in main
    run(vars(opt))
  File "/home/jarvis/anaconda3/envs/py38/lib/python3.8/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/examples/track.py", line 126, in run
    predictor.tracker_outputs[i] = predictor.trackers[i].update(dets.cpu().detach().numpy(), im0)
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/boxmot/strongsort/strong_sort.py", line 64, in update
    self.tracker.update(detections, clss, confs)
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/boxmot/strongsort/sort/tracker.py", line 95, in update
    self.tracks[track_idx].update(
  File "/home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking/boxmot/strongsort/sort/track.py", line 279, in update
    self.class_id = class_id.int()
AttributeError: 'numpy.float32' object has no attribute 'int'

2023-06-15 16:56:35.030 | ERROR    | val:eval:224 - Successfully loaded imagenet pretrained weights from "/home/jarvis/.cache/torch/checkpoints/osnet_x1_0_imagenet.pth"

(py38) jarvis@jia:~/coding/pyhome/mikel-brostrom/yolo_tracking$ python -V
Python 3.8.16
(py38) jarvis@jia:~/coding/pyhome/mikel-brostrom/yolo_tracking$ pip list | grep boxmot
boxmot                        10.0.13      /home/jarvis/coding/pyhome/mikel-brostrom/yolo_tracking
(py38) jarvis@jia:~/coding/pyhome/mikel-brostrom/yolo_tracking$ 

[meua]

Problem solved, edit boxmot/strongsort/sort/track.py line 279,

-        self.class_id = class_id.int()
+        self.class_id = np.int32(class_id)

[mikel-brostrom]

Problem solved, edit boxmot/strongsort/sort/track.py line 279,

-        self.class_id = class_id.int()
+        self.class_id = np.int32(class_id)

Fixed based on this. Turned this variable into an INT64 for larger ID range