robmarkcole/fire-detection-from-images

Use smaller yolov5

Opened this issue · 1 comments

robmarkcole commented

Have been training with --img 416 but images are approx 285x180 pixels, meaning I have been oversized. This means the generated anchor boxes are probably also oversized

robmarkcole commented

Noticed that roboflow actually stretches to 416x416 as a preprocessing step, so this might not be an issue