Ayomide
26 lines
801 B
Markdown
26 lines
801 B
Markdown
**1. Algorithm Choice**
|
|
- **Selected:** YOLOv8n (lightweight version)
|
|
- **Why:**
|
|
- Fast detection (0.5s/image on CPU)
|
|
- Works well with small datasets (40 images)
|
|
- Accurate for motherboard components
|
|
|
|
**2. Hardware Impact**
|
|
- **Training:**
|
|
- GPU recommended (4x faster training)
|
|
- CPU works but slower
|
|
- **Deployment:**
|
|
- CPU sufficient for basic use
|
|
- GPU better for high volume
|
|
|
|
**3. Video Handling**
|
|
- **Approach:** Process each frame individually
|
|
- **Changes Needed:**
|
|
- Add frame-by-frame processing
|
|
- Include tracking to follow memory modules
|
|
- Optimize for speed (lower resolution helps)
|
|
|
|
**Key Facts:**
|
|
- Same model works for images/video
|
|
- CPU processing is practical
|
|
- No architecture changes needed between image/video modes. |