CFENet: Contextual Feature Enhancement Network for Tiny Object Detection in Aerial Images

With the development of deep learning techniques and object detectors, the performance of object detection has been rapidly improved. However, since tiny objects contain only a small number of pixels and lack appearance information, this creates difficulties for detector recognition. Although existing research has improved detection performance by fusing different feature layers to enhance feature information of objects, this also leads to the problem of mixed feature information, especially for tiny objects where features are easily covered, which exacerbates the difficulty of recognition. To solve the above problems, we propose a contextual feature enhancement network (CFENet), which is an efficient framework built on anchor-based object detectors. In CFENet, to effectively utilize contextual information around an object to enhance the detection of tiny objects, we use poolFormer to build a backbone to extract object features. To alleviate the feature blending problem caused by feature fusion, we propose a feature suppression module (FSM) that effectively suppresses background information and redundant features to enhance tiny object features. In addition, we utilize the improved Gaussian Wasserstein distance loss to modify the loss function to obtain high-quality bounding boxes, and we further manipulate the shallow feature layer of the output and then add a detection head to enhance the detection of tiny objects. We have conducted extensive experiments on the public datasets AI-TOD, VisDrone, and DOTA to demonstrate the effectiveness of our approach.

I. Introduction

With the development of deep learning techniques, the performance of object detection has rapidly improved in a short period of time. In contrast to small, medium, and large objects defined in the MS COCO benchmark [1], the AI-TOD benchmark [2] defines tiny objects as those with dimensions smaller than pixels. However, compared with the detection of larger objects, small objects themselves have fewer pixels in their extremely small size, weaker feature rendering ability, and their appearance information is extremely limited, resulting in less spatial information, and tiny objects are easily confused with the background, increasing background interference [3], [4]. Due to the limitation of spatial resolution, it is difficult for conventional object detectors to effectively capture the features required for accurate classification and localization of tiny objects, resulting in a large number of failures in detecting tiny objects [2], [5], [6], [7].