MTU-Net: Multilevel TransUNet for Space-Based Infrared Tiny Ship Detection

Abstract

Space-based infrared tiny ship detection aims at separating tiny ships from the images captured by earth orbiting satellites. Due to the extremely large image coverage area (e.g., thousands square kilometers), candidate targets in these images are much smaller, dimer, more changeable than those targets observed by aerial-based and land-based imaging devices. Existing short imaging distance-based infrared datasets and target detection methods cannot be well adopted to the space-based surveillance task. To address these problems, we develop a spacebased infrared tiny ship detection dataset (namely, NUDT-SIRST-Sea) with 48 space-based infrared images and 17598 pixel-level tiny ship annotations. Each image covers about 10000 square kilometers of area with 10000 × 10000 pixels. Considering the extreme characteristics (e.g., small, dim, changeable) of those tiny ships in such challenging scenes, we propose a multi-level TransUNet (MTU-Net) in this paper. Specifically, we design a Vision Transformer (ViT) Convolutional Neural Network (CNN) hybrid encoder to extract multi-level features. Local feature maps are first extracted by several convolution layers and then fed into the multi-level feature extraction module (MVTM) to capture long-distance dependency. We further propose a copy-rotateresize-paste (CRRP) data augmentation approach to accelerate the training phase, which effectively alleviates the issue of sample imbalance between targets and background. Besides, we design a FocalIoU loss to achieve both target localization and shape description. Experimental results on the NUDT-SIRST-Sea dataset show that our MTU-Net outperforms traditional and existing deep learning based SIRST methods in terms of probability of detection, false alarm rate and intersection over union. Our code is available at: https://github.com/TianhaoWu16/Multi-level-TransUNet-for-Space-based-Infrared-Tiny-ship-Detection.

DEAR: A Novel Deep-Level Semantics Feature Reinforce Framework for Infrared Small Object Segmentation

“…To better extract knowledge and patterns of small objects from different types of infrared data, data-driven neural network methods have emerged and received extensive research attention. Early methods employed encoder-decoder structures with the expectation of accurately identifying small objects [14,17,18]. However, due to the limited information content of small objects in images, they are prone to being overwhelmed by the background during decoder reconstruction.…”

LW-IRSTNet: Lightweight Infrared Small Target Segmentation Network and Application Deployment

“…They believed that existing deep-learning-based methods were limited by the locality of CNNs, weakening their ability to capture large-range dependencies. Wu et al [33] proposed the multi-level TransUNet (MTUNet), which uses a hybrid ViT encoder and CNN to extract multi-level features.…”

“…2) Compare baselines. U-Net [21], FusionNet [53], ENet [37], BiSeNet [45], DeeplabV3+ [50], DABNet [54], DFANet [46], CGNet [55], ACM-UNet [19], LSPMNet [25], DNANet [24], AGPCNet [22], ABCNet [56], MTUNet [33], and ours (LW-IRSTNet). Fusion datasets (MDFA [18] (Black hot) + SIRST [19] (White hot)+ SIRST Aug [22] (White hot)+ NUDT-SIRST [24] 3) Evaluation metrics.…”

Multiscale Feature-Based Infrared Ship Detection

“…In recent years, researchers have been exploring various methods for target detection in complex sea scenarios. The existing methods can be broadly divided into the following four categories: background-suppression based [6][7][8][9], active-contour based [10,11], featureanalysis based [12][13][14], and deep learning [15][16][17]. The background-suppression based method assumes that the grayscale distribution of the sea background is stable so that the modeling and estimation of the sea background can be achieved using different statistical models.…”