An Augmentation Small Object Detection Method Based on NAS-FPN

时间:2020-11-25 19:07:51

To solve the problem that small object detection accuracy is poor caused by fewer pixels contained and difficult feature extraction and improving pyramidal feature re-presentations, this paper proposes an effective method of enhanced small detection based on a new feature pyramid architecture adopting Neural Architecture Search (NAS-FPN). We adopt scaling transformation, contrast enhancement, flipping, brightness alteration and rotate with a random angle as our augmentation approach to process the input data for better small object detection effect. As a result, the accuracy of our methods tested on VOC2007 test set is relatively higher than FPN and NAS-FPN, especially on small object detection such as boat, bottle and chair, as presented in TABLE.II.