GoG rpi5 and ncnn model

[ElectroS3B]

I try to use the owl ultralytics-gog branch but I have a problem, here is the error :

WARNING ⚠️ Unable to automatically guess model task, assuming 'task=detect'. Explicitly define task for your model, i.e. 'task=detect', 'segment', 'classify','pose' or 'obb'.
Nozzle 1 Nozzle 2 Nozzle 3 Nozzle 4
Loading models/best_ncnn_model for NCNN inference...
2025-02-16 22:33:04 - ERROR - [main] - [CRITICAL ERROR] STOPPED: cannot access local variable 'image_out' where it is not associated with a value
2025-02-16 22:33:04 - INFO - [picamera2.picamera2] - Camera stopped
2025-02-16 22:33:07 - INFO - [picamera2.picamera2] - Camera closed successfully.
(owl) owl@raspberrypi:~/owl $

[geezacoleman]

I can't seem to replicate this - were you able to resolve it? Can you provide your config file/settings?

The warning isn't an issue, it assumes correctly that the task=detect. If image_out not having a value implies that the weed_detector.inference() with the GoG algorithm, is not producing an image.