Scale Enhancement Pyramid Network for Small Object Detection from UAV Images

Abstract

Object detection is challenging in large-scale images captured by unmanned aerial vehicles (UAVs), especially when detecting small objects with significant scale variation. Most solutions employ the fusion of different scale features by building multi-scale feature pyramids to ensure that the detail and semantic information are abundant. Although feature fusion benefits object detection, it still requires the long-range dependencies information necessary for small objects with significant scale variation detection. We propose a simple yet effective scale enhancement pyramid network (SEPNet) to address these problems. A SEPNet consists of a context enhancement module (CEM) and feature alignment module (FAM). Technically, the CEM combines multi-scale atrous convolution and multi-branch grouped convolution to model global relationships. Additionally, it enhances object feature representation, preventing features with lost spatial information from flowing into the feature pyramid network (FPN). The FAM adaptively learns offsets of pixels to preserve feature consistency. The FAM aims to adjust the location of sampling points in the convolutional kernel, effectively alleviating information conflict caused by the fusion of adjacent features. Results indicate that the SEPNet achieves an AP score of 18.9% on VisDrone, which is 7.1% higher than the AP score of state-of-the-art detectors RetinaNet achieves an AP score of 81.5% on PASCAL VOC.

Overview of Image Datasets for Deep Learning Applications in Diagnostics of Power Infrastructure

“…The latter use of them has also been developed. As shown in [ 26 , 38 , 40 , 42 , 43 , 46 , 51 , 52 ], this type of imagery can be used for many tasks that support the proper operation of the network infrastructure, which usually operates over a large area.…”

“…Applications that use images collected by UAVs are also popular because they allow the development of a well-performing algorithm. Another example is the scale enhancement pyramid network, which achieved an AP score of 81.5% for small object detection [ 52 ]. It is also due to the fact that one of the most practical and certainly popular ways of image acquisition is the use of unmanned aerial vehicles (UAVs).…”

“…It is also due to the fact that one of the most practical and certainly popular ways of image acquisition is the use of unmanned aerial vehicles (UAVs). Such devices have been used for the preparation of datasets, e.g., [ 26 , 40 , 42 , 43 , 44 , 46 , 48 , 51 , 52 ]. Among other things, they helped to provide a power infrastructure classification model using deep neural networks that achieved an overall F-score of 75% for multi-class classification scenarios and 88% for pylon identification [ 46 ].…”

Weather-Domain Transfer-Based Attention YOLO for Multi-Domain Insulator Defect Detection and Classification in UAV Images

“…With the continuous development of UAV technology [18], UAV inspection has been widely used in various industries. Various weather conditions can cause damage to the insulators of power transmission lines, even leading to failures [19].…”

Enhanced Infrared Detection Algorithm for Weak Targets in Complex Backgrounds

“…False alarm rate, Fa [24]: The result is obtained by calculating the ratio of false prediction pixels to all pixels in the image, where P f is the false prediction pixel and Pa is all pixels in the image:…”