We intuitively know poor quality images lead to poor object detection and vice-versa.
Image-Adaptive (IA)-YOLO, published 3 days ago, uses 2 models in a hybrid task:
- One model (trained in weakly supervised manner) that learns how to adaptively improve an image,
- and a YOLOv3 model to train using the improved images
The authors show some impressive results.
This idea could also be extended to classify poor quality images.
