Work Accomplished:
- Went to Aldis to test out hypothetical workflow
- Collected additional Aldi product images from GitHub repositories and online sources to refine detection capabilities.
- Researched fine-tuning techniques to adapt YOLOv8 to Aldi-specific products, including augmentation strategies to improve model robustness.
- Began implementing an OpenCV preprocessing pipeline for contrast adjustment and text detection (OCR) to assist in product differentiation.
Progress Status:
I am making steady progress in adapting YOLOv8 for real-time grocery detection. While I have successfully set up the base model and tested inference speed, the next step is fine-tuning it with Aldi-specific images and integrating OCR-based product matching.
Next Weeks Goal:
- Implementing OCR-based product matching, testing label extraction on Aldi-specific items.
- Integrating detected objects with the product database, ensuring proper matching and categorization.