Translated Project Presentation (Eng)

Slide 15 - Hardware Raspberry Pi 4: 4 GB RAM, 4 cores, Linux support, low power consumption Raspberry Pi Camera V2: 8 MP, 1080p video Computing power, low consumption and connectivity. Slide 16 - ⚠️ Power Supply Limitation However, during testing: The camera operated unstably due to power supply issues. This limited real-time usage and compromised performance. 🔄 Alternative Solution Android Application "IP Webcam" Slide 17 - Operating Procedure best.pt → Raspberry Pi → IP Webcam → Python Inference [Specs: ARM Cortex-A72, 4 GB RAM] → Approx. 1 FPS (985 ms/image) Slide 18 - Errors Encountered ⚠️ Error --> ✅ Solution Colab runtime disconnected --> save checkpoint CUDA unavailable (GPU not enabled) --> --device cpu or enable GPU in runtime Cross-dependency conflicts --> Downgrade versions to avoid conflict Dataset not found (relative paths) --> Switch to absolute paths in data.yaml Slide 19 - Maintenance This fits within a predictive maintenance logic — detecting defects before they cause a breakdown or quality loss. Integration of a corrective maintenance function, in the form of contextual pop-up messages that automatically appear for each type of detected defect. Pop-up example: 🔴 Defect detected: Open_circuit ✅ Recommended action: Re-solder the connection using a repair wire.