UEFPN: Unified and Enhanced Feature Pyramid Networks for Small Object Detection

What is it about?

Object detection models based on feature pyramid networks have made signiicant progress in general object detection. However, small object detection is still a challenge for the existing models. In this paper, we think that two factors in the existing feature pyramid networks inhibit the performance of small object detection. The irst one is that the diferent feature domains of shallow and deep layer features inhibit the model performance. The second one is that the accumulation of upper layer features leads to feature aliasing efect on the lower layer features, which interferes with the representations of small object features. Therefore, we propose Uniied and Enhanced Feature Pyramid Networks (UEFPN) to improve the APs and ARs of small object detection. It has the following three characteristics: (1) Using the deep features of high-resolution image and original image to form the multi-scale features of uniied domain. (2) In multi-scale features fusion, we learn the importance of upper layer features with the Channel Attention Fusion module (CAF), to optimize feature aliasing efect and enhance the context information of shallow layer features. (3) UEFPN can be quickly applied to diferent models. The results of lots of experiments show that the models with UEFPN achieve signiicant performance improvement in small object detection compared with the baseline models.

Featured Image

Photo by William Bayreuther on Unsplash

Photo by William Bayreuther on Unsplash