search

roboflow / notebooks

Examples and tutorials on using SOTA computer vision models and techniques. Learn everything from old-school ResNet, through YOLO and object-detection transformers like DETR, to the latest models like Grounding DINO and SAM.
https://roboflow.com/models
5.58k stars 880 forks source link

_pickle.UnpicklingError: invalid load key, '<'. #307

Closed aamirali13 closed 2 months ago

aamirali13 commented 2 months ago

Search before asking

Notebook name

Track and Count Vehicles with YOLOv8 + ByteTRACK + Supervision

Bug

I was trying to run the notebook by using my own weights but getting the mentioned error throughout. It's not working for even the weights which are shared with link.

Environment

Google colab

Minimal Reproducible Example

!python detect.py --weights content/best.pt --img 1280 --conf 0.25 --source content/08fd33_4.mp4 --name custom

or

model = torch.hub.load('ultralytics/yolov5', 'custom', WEIGHTS_PATH, device=0,force_reload=True)

Additional

No response

Are you willing to submit a PR?

LinasKo commented 2 months ago

Hi @aamirali13 👋

The notebook you mentioned uses Yolov8, whereas the code you pasted mentions yolov5. These two models are incompatible - maybe that's related to your issue?

aamirali13 commented 2 months ago

Sorry for the misunderstanding. Notebook : Football Players Tracking with YOLOv5 + ByteTRACK

This is what I am trying to run and getting issue.

LinasKo commented 2 months ago

Hi @aamirali13, unfortunately, figuring out the issue is taking more time than I have to spare.

Instead, I'd like to offer an alternative. Instead of this 2-year-old notebook, here's a recent project by my colleague, digging very deep into the problem of tracking in football. Unlike this notebook, it uses recent models and supervision features.

While there's a little bit more info to pick through, I bet you'll find it both more performant and easier to use (that's what we've been working towards!).

Alternatively, if you're looking for a quick way to track objects, check out our docs: https://supervision.roboflow.com/latest/how_to/track_objects/

Best of luck!

aamirali13 commented 2 months ago

Ok Thanks @LinasKo I'll definitely look into your suggested resource