With the development of infrared technology, infrared dim and small target detection plays an essential role in precision guidance and early warning systems. Due to the low contrast and signal-to-noise ratio that characterizes infrared dim and small target in images, the dim and small target can easily be drowned out by noise and background. A new infrared dim and small target detection network (CA-U2-Net) is proposed to address the challenge of infrared weak target detection and shape retention in complex backgrounds. Specifically, firstly, the U2-Net network structure has been improved to prevent the loss of shallow information due to increased network depth and to make it more suitable for detecting the dim and small target. Then, the upper and lower attention module was designed on the network to make the model more focused on dim and small target features while suppressing irrelevant information, further improving the detection rate. Finally, a contour detection branch was added to the top of the model to fuse the contour detection map with the feature map to get a better target shape. After experimental evaluation, the method achieved a detection rate of 97.17% and retained a more accurate infrared dim and small target shape. Compared to other advanced methods, our method performs better in detection rate, false detection rate and shape retention. In addition, a new infrared dim and small target dataset consisting of 10,000 images was constructed.

With the development of infrared technology, infrared dim and small target detection plays a vital role in precision guidance applications. To address the problems of insufficient dataset coverage and huge actual shooting costs in infrared dim and small target detection methods, this paper proposes a method for generating infrared dim and small target sequence datasets based on generative adversarial networks (GANs). Specifically, first, the improved deep convolutional generative adversarial network (DCGAN) model is used to generate clear images of the infrared sky background. Then, target–background sequence images are constructed using multi-scale feature extraction and improved conditional generative adversarial networks. This method fully considers the infrared characteristics of the target and the background, which can achieve effective expansion of the image data and provide a test set for the infrared small target detection and recognition algorithm. In addition, the classifier’s performance can be improved by expanding the training set, which enhances the accuracy and effect of infrared dim and small target detection based on deep learning. After experimental evaluation, the dataset generated by this method is similar to the real infrared dataset, and the model detection accuracy can be improved after training with the latest deep learning model.

The rapid development of infrared technology makes it widely applicable in such fields as military, medical, testing and communications.This paper proposes an algorithm based on local contrast and gradient (LCG). Specifically, the algorithm uses the difference of gaussian and threshold segmentation to preprocess to obtain the position of possible target points. Then, local contrast processing is conducted for the possible target points. Finally, the global is processed using an improved directional gradient. According to the experimental results, the proposed algorithm outperforms the existing algorithms in terms of detection probability and false detection probability.

Recent grid-based document representations like BERTgrid allow the simultaneous encoding of the textual and layout information of a document in a 2D feature map so that state-of-the-art image segmentation and/or object detection models can be straightforwardly leveraged to extract key information from documents. However, such methods have not achieved comparable performance to state-of-the-art sequence- and graph-based methods such as LayoutLM and PICK yet. In this paper, we propose a new multi-modal backbone network by concatenating a BERTgrid to an intermediate layer of a CNN model, where the input of CNN is a document image and the BERTgrid is a grid of word embeddings, to generate a more powerful grid-based document representation, named ViBERTgrid. Unlike BERTgrid, the parameters of BERT and CNN in our multimodal backbone network are trained jointly. Our experimental results demonstrate that this joint training strategy improves significantly the representation ability of ViBERTgrid. Consequently, our ViBERTgrid-based key information extraction approach has achieved state-of-the-art performance on real-world datasets.