Abstract:Infrared small target detection has always been a difficult problem in the field of object detection. The main reason affecting the accuracy is that the small infrared target has fewer pixels and weaker features. The current optimization methods for the small target are mainly based on multi-scale feature fusion or super-resolution enhancement. When super-resolution networks are applied to infrared target detection, there are still two non-negligible problems: first, the super-resolution structure will consume too much arithmetic power, resulting in a low detection rate, and second, the low-resolution images characterizing small targets are usually obtained by downsampling with high-resolution images during training, which is different from the distribution of tiny target in actual detection applications, resulting in poor detection accuracy. We propose a new detection network to solve the above problem: Region Super Resolution Generative Adversarial Network(RSRGAN). It contains a simple structured network, Region Context Network(RCN) as the backbone that consumes less computational cost to extract the possible regions. The generator of the Generative Adversarial Network(GAN) includes two modules: distribution transformation and super-resolution enhancement. First, the blurred infrared small target is converted into a clear target with similar distribution as the training set. Then the resolution is increased, which can achieve a better enhancement effect. The discriminator distinguishes whether the input comes from the generator or the actual image to assist in generating a better super-resolution image. Meanwhile, we produced an infrared Unmanned Aerial Vehicle(UAV) small target dataset, target pixels below 20*20, containing birds, leaves, and other similar disturbances, which is more challenging for the detection algorithm. Our method proves better detection of small IR targets and shows superior performance over state-of-the-art methods through experiments.

Small Object Detection Leveraging on Simultaneous Super-resolution.

Compressive Sensing Ghost Imaging Object Detection Using Generative Adversarial Networks

Small Object Detection in Remote Sensing Images with Residual Feature Aggregation-Based Super-Resolution and Object Detector Network

Perceptual Generative Adversarial Networks for Small Object Detection

Multi-task Generative Adversarial Network for Detecting Small Objects in the Wild

Small Object Detection Networks Based on Classification-Oriented Super-Resolution GAN for UAV Aerial Imagery

Centered Multi-Task Generative Adversarial Network for Small Object Detection

Single Remote Sensing Image Super-Resolution Via a Generative Adversarial Network with Stratified Dense Sampling and Chain Training

Small object detection model for remote sensing images combining super-resolution assisted reasoning and dynamic feature fusion

Small-Object Detection in Remote Sensing Images With Super-Resolution Perception

Small-Object Detection Based on YOLO and Dense Block via Image Super-Resolution

Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

SODSR: A Three-Stage Small Object Detection Via Super-Resolution Using Optimizing Combination

Small Object Detection by Generative and Discriminative Learning.

SuperYOLO: Super Resolution Assisted Object Detection in Multimodal Remote Sensing Imagery

Target detection algorithm based on super- resolution color remote sensing image reconstruction

Infrared small target detection via region super resolution generative adversarial network

One-Stage Small Object Detection Using Super-Resolved Feature Map for Edge Devices

Small Object Detection Method Based on Adaptive Spatial Parallel Convolution and Fast Multi-Scale Fusion

Small object detection in remote sensing images based on super-resolution

Super-Resolution GAN and Global Aware Object Detection System for Vehicle Detection in Complex Traffic Environments