Torch.hub.load() broken because of a recent commit

[chainyo]

🐛 Bug

Since this commit 5ed2860 we can't load a custom model via torch.hub.load() function, because the yolov5 repo is by default cached and isn't downloaded in the project repository.

These are the lines that changed in multiple files

- ROOT = ROOT.relative_to(Path.cwd()) if ROOT.is_relative_to(Path.cwd()) else ROOT # relative

+ ROOT = ROOT.relative_to(Path.cwd()) # relative

To Reproduce (REQUIRED)

Input:

import torch

model = torch.hub.load("ultralytics/yolov5", "custom", path="path/to/yolov5.pt", force_reload=True)

(I can't include my custom model because it's my company property, but any .pt will do the job!)

Output:

Downloading: "https://github.com/ultralytics/yolov5/archive/master.zip" to /home/Thomas.Chaigneau/.cache/torch/hub/master.zip
Traceback (most recent call last):
  File "/home/Thomas.Chaigneau/miniconda3/envs/project/lib/python3.8/site-packages/project/services/yolov5service.py", line 23, in __init__
    model = torch.hub.load(
  File "/home/Thomas.Chaigneau/.local/lib/python3.8/site-packages/torch/hub.py", line 364, in load
    model = _load_local(repo_or_dir, model, *args, **kwargs)
  File "/home/Thomas.Chaigneau/.local/lib/python3.8/site-packages/torch/hub.py", line 393, in _load_local
    model = entry(*args, **kwargs)
  File "/home/Thomas.Chaigneau/.cache/torch/hub/ultralytics_yolov5_master/hubconf.py", line 70, in custom
    return _create(path, autoshape=autoshape, verbose=verbose, device=device)
  File "/home/Thomas.Chaigneau/.cache/torch/hub/ultralytics_yolov5_master/hubconf.py", line 30, in _create
    from models.yolo import Model
  File "/home/Thomas.Chaigneau/.cache/torch/hub/ultralytics_yolov5_master/models/yolo.py", line 18, in <module>
    ROOT = ROOT.relative_to(Path.cwd())  # relative
  File "/home/Thomas.Chaigneau/miniconda3/envs/project/lib/python3.8/pathlib.py", line 908, in relative_to
    raise ValueError("{!r} does not start with {!r}"
ValueError: '/home/Thomas.Chaigneau/.cache/torch/hub/ultralytics_yolov5_master' does not start with '/home/Thomas.Chaigneau/code/project'

Expected behavior

This worked the day before.
I found a workaround here #4970 but it's only temporary :)

[github-actions]

👋 Hello @chainyo, thank you for your interest in YOLOv5 🚀! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://ultralytics.com or email Glenn Jocher at glenn.jocher@ultralytics.com.

Requirements

Python>=3.6.0 with all requirements.txt installed including PyTorch>=1.7. To get started:

$ git clone https://github.com/ultralytics/yolov5
$ cd yolov5
$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), validation (val.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

@chainyo good news 😃! Your original issue may now be fixed ✅ in PR #4974. This PR comments out relative ROOT attempts in yolo.py and tf.py which may be imported by PyTorch Hub. We did not revert to using Path().is_relative() anymore because this was only introduced in python 3.8 and we want python >3.6 compatability.

To receive this update:

• Git – git pull from within your yolov5/ directory or git clone https://github.com/ultralytics/yolov5 again
• PyTorch Hub – Force-reload with model = torch.hub.load('ultralytics/yolov5', 'yolov5s', force_reload=True)
• Notebooks – View updated notebooks
• Docker – sudo docker pull ultralytics/yolov5:latest to update your image

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 🚀!

[chainyo]

Thank you for spotting this issue and informing us of the problem. Please let us know if this update resolves the issue for you, and feel free to inform us of any other issues you discover or feature requests that come to mind. Happy trainings with YOLOv5 rocket!

Yes problem fixed! Hope you will find a way to make it work without commenting the line you wanted to implement
Thanks for the fast answer, have a nice day ☀️ !

[glenn-jocher]

@chainyo thanks for the confirmation!

Yes there is an alternate os.path-based solution that we can implement later. The only reason for updating ROOT to a relative path is to reduce screen printout on long paths, but these updates are not critical to functionality.