Issue Exporting Model to ONNX Format - TypeError in check_imgsz

[Lawrence-Leung]

Hello,

I've been working with your repository and followed the instructions in the README.md to set up the environment. My Python and PyTorch versions are aligned with what's specified in the README.md.

I encountered an issue while attempting to export a YOLOv8 model to ONNX format. Below is the sequence of commands I used:

import torch
import sys

# Temporarily adding the modified Ultralytics to PATH
sys.path.insert(0, "/home/lawrence/projects/Phytium2024-Local/other_repos/YOLOv8-multi-task/ultalytics")

from ultralytics import YOLO
model = YOLO('/home/lawrence/projects/Phytium2024-Local/other_repos/YOLOv8-multi-task/ultralytics/models/v8/yolov8-bdd-v4-one-dropout-individual-n.yaml', task='multi').load('/home/lawrence/projects/Phytium2024-Local/other_repos/v4.pt')
model.info()

# Attempt to export the model to ONNX format
model.export(format='onnx', half=False, dynamic=False, opset=12)

However, I encountered the following issue during the export process:

Ultralytics YOLOv8.0.105 🚀 Python-3.9.18 torch-2.2.1+cu121 CPU
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/lawrence/projects/Phytium2024-Local/other_repos/YOLOv8-multi-task/ultralytics/yolo/engine/model.py", line 356, in export
    return Exporter(overrides=args, _callbacks=self.callbacks)(model=self.model)
  File "/opt/anaconda3/lib/python3.9/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/home/lawrence/projects/Phytium2024-Local/other_repos/YOLOv8-multi-task/ultralytics/yolo/engine/exporter.py", line 167, in __call__
    self.imgsz = check_imgsz(self.args.imgsz, stride=model.stride, min_dim=2)  # check image size
  File "/home/lawrence/projects/Phytium2024-Local/other_repos/YOLOv8-multi-task/ultralytics/yolo/utils/checks.py", line 59, in check_imgsz
    stride = int(stride.max() if isinstance(stride, torch.Tensor) else stride)
TypeError: int() argument must be a string, a bytes-like object or a number, not 'list'

Could you please help me understand what might be causing this issue and how I can resolve it to successfully export my model to ONNX format?

Thank you for your time and assistance.

[Lawrence-Leung]

There's something to be added that the model that I've mentioned is based on yolov8-bdd-v4-one-dropout-individual-n.yaml, which is located at directory /ultralytics/models/v8 relative to the root directory of your repository.

[loganwu0526]

hi, @Lawrence-Leung , do you solve this issue? I also have this issue when I export my model to onnx

thanks

[1623021453]

hi, @Lawrence-Leung , do you solve this issue? I also have this issue when I export my model to onnx

thanks

[Lawrence-Leung]

Hi, @JiayuanWang-JW, Thank you for your reply, but I'm so sorry that I failed to solve this issue, since my team member and I rejected using your solution into our project on a meeting in March 2024, for we think that instead of using multi-task on the level of the inference model, it's much better to put multi-task on the level of processes of operating system(i.e. Linux kernel), which is less experimental but more stable for robustness of our application. Our recent work is not a big project though, as we are just undergraduate students busy with curriculum studies, and most importantly, thriving for entrance of postgraduate studies in a better university. My team members and I are participating in a competition of electronic engineering for university students throughout China, called Phytium Cup (飞腾杯), which is a branch of a series called Integrated Circuit Innovation and Entrepreneurship Competition (集成电路创新创业大赛, where I'm not sure this translation is accurate or not). Our goal is to make a robust application on hardwares provided by Phytium company, especially the performance on AMP(asymmetric multi-processing). Though we're unable to do anything for improving your project, we still sincerely thank you for your attention to this issue. Best regards, Lawrence 2024年7月16日上午11:06，1623021453 @.> 写道： hi, @Lawrence-Leung , do you solve this issue? I also have this issue when I export my model to onnx thanks — Reply to this email directly, view it on GitHub , or unsubscribe . You are receiving this because you were mentioned. Message ID: @.>