YOLOv10 on UNIHIKER
Matha.Goram 2024-08-31 18:09:26 202 Views1 Replies While stepping through the tutorial at https://www.dfrobot.com/blog-13928.html, I encountered the following error message:
yolo) root@unihiker1:~/projects/yolov10# python3 ./quick_start.py
Illegal instruction
(yolo) root@unihiker1:~/projects/yolov10#
I completed all the preceding steps in the tutorial except:
3 Quick Start: Running Native YOLOv10
Step 2: Prepare the required file
[yolov10.pt]
What are the instructions to create this file?
I'm using the Python script file without any changes, and I did clone the repository as instructed in the Tutorial.
All preceding steps for the installation completed without any error message or warning. I know I should be reading the documentation in greater detail, but if someone could guide me with this basic issue (pt file missing?), I would appreciate it very much. Thanks.
Regards.
Update:
The documentation at https://www.dfrobot.com/blog-13928.html at Step 5 Run the Python script states that the quick_start.py script (created at an earlier step) should be run. The image immediately following this statement illustrates the output from a script call test_onnx_bus.py.
Also, there is some (minor difference, IMHO) between the statements in quick_start.py and the jameslahm/yolov10 example. The DFR statement is implicit for v10 while the equivalent statement at jameslahm website is syntactically different.
In the DFR tutorial, the model loading is incomplete owing to the Illegal instruction message. It would be helpful if the person at DFR who wrote the tutorial could offer some guidance. Thanks.
Regards.
P.S.
I'm curious about having to install huggingface but not importing it explicitly in the simple Python quick_start.py script, whereas jameslahm does it slightly differently:
at https://huggingface.co/jameslahm/yolov10n
Training and validation
from ultralytics import YOLOv10
model = YOLOv10.from_pretrained('jameslahm/yolov10n')
# Training
model.train(...)
# after training, one can push to the hub
model.push_to_hub("your-hf-username/yolov10-finetuned")
# Validation
model.val(...)
Inference
from ultralytics import YOLOv10
model = YOLOv10.from_pretrained('jameslahm/yolov10n')
source = 'http://images.cocodataset.org/val2017/000000039769.jpg'
model.predict(source=source, save=True)
Matha.Goram 
