Target detection based on deep learning are developing rapidly. However, small target detection is still a challenge. In this paper, a simple and efficient network for small target detection is proposed. We put forward to improve the detection performance of the small targets in three aspects. Firstly, as the contextual information is important to detect the small targets, we proposed to use ‘dilated module’ to expand the receptive field without loss of resolution or coverage. Secondly, we applied feature fusion in different dilated modules to improve the ability of network in detecting small targets. Finally, we used ‘passthrough module’ to get the finer-grained information from the earlier layer and combined it with the semantic information from the deeper layer. To improve the detection speed of the network, it is proposed to use 1×1 convolution to reduce the dimension of the network. We composed small vehicle dataset based on VEDAI dataset and DOTA dataset respectively and also analyzed the distribution of the small targets in each dataset. To evaluate the performance of the proposed network, we trained the model on the dataset above and compared with the state-of-the-art target detection algorithms, our approach achieved 80.16% average precision (AP) on VEDAI dataset and 88.63% AP on DOTA dataset and the frames per second (FPS) is 75.4. The AP of our network is much better than the result of tiny YOLO V3 and is nearly the same as the result of YOLO V3. However, the FPS of our network is almost the same as that of tiny YOLO V3.

Object detection is a fundamental capability that enables drones to perform various tasks. However, achieving a suitable equilibrium between performance, efficiency, and lightweight design continues to be a significant challenge for current algorithms. To address this issue, we propose an enhanced small object detection transformer model called ESO-DETR. First, we present a gated single-head attention backbone block, known as the GSHA block, which enhances the extraction of local details. Besides, ESO-DETR utilizes the multiscale multihead self-attention mechanism (MMSA) to efficiently manage complex features within its backbone network. We also introduce a novel and efficient feature fusion pyramid network for enhanced small object detection, termed ESO-FPN. This network integrates large convolutional kernels with dual-domain attention mechanisms. Lastly, we introduce the EMASlideVariFocal loss (ESVF Loss), which dynamically adjusts the weights to improve the model’s focus on more challenging samples. In comparison with the baseline model, ESO-DETR demonstrates enhancements of 3.9% and 4.0% in the $mA{P}_{50}$ metric on the VisDrone and HIT-UAV datasets, respectively, while also reducing parameters by 25%. These results highlight the capability of ESO-DETR to improve detection accuracy while maintaining a lightweight and efficient structure.

Pedestrian flow statistics and analysis in public places is an important means to ensure urban safety. However, in recent years, a video-based pedestrian flow statistics algorithm mainly relies on binocular vision or a vertical downward camera, which has serious limitations on the application scene and counting area, and cannot make use of the large number of monocular cameras in the city. To solve this problem, we propose a pedestrian flow statistics algorithm based on monocular camera. Firstly, a convolution neural network is used to detect the pedestrian targets. Then, with a Kalman filter, the motion models for the targets are established. Based on these motion models, data association algorithm completes target tracking. Finally, the pedestrian flow is counted by the pedestrian counting method based on virtual blocks. The algorithm is tested on real scenes and public data sets. The experimental results show that the algorithm has high accuracy and strong real-time performance, which verifies the reliability of the algorithm.

For many pedestrian detectors, background vs. foreground errors heavily influence the detection quality. Our main contribution is to design semantic regions of interest that extract the foreground target roughly to reduce the background vs. foreground errors of detectors. First, we generate a pedestrian heat map from the input image with a full convolutional neural network trained on the Caltech Pedestrian Dataset. Next, semantic regions of interest are extracted from the heat map by morphological image processing. Finally, the semantic regions of interest divide the whole image into foreground and background to assist the decision-making of detectors. We test our approach on the Caltech Pedestrian Detection Benchmark. With the help of our semantic regions of interest, the effects of the detectors have varying degrees of improvement. The best one exceeds the state-of-the-art.

In recent years, multi-object tracking has attracted more and more attention, both in academia and engineering, but most of the recent works do not pay attention to the speed of the algorithm, and only pursue the accuracy. In this article, we propose an online multi-pedestrian tracking algorithm, taking into account both the accuracy and the speed. Firstly, the motion models of the targets are established by Kalman filter. At the same time, the appearance models of the targets are extracted by convolutional neural network. Moreover, a data association algorithm is proposed, which integrates the motion information, including scale, Intersection-over-Union and distance, and the appearance information, including the current appearance model and the long-term appearance model. With the data association algorithm, the matching between detections and tracklets is realized, and the goal of tracking by detection is achieved. We compare the proposed algorithm with other algorithms on the MOT15 benchmark and the MOT16 benchmark. The experiment results show that the algorithm has high accuracy and good real-time performance.

In this article, we aim to improve the performance of synthetic aperture radar (SAR) ship detection under complex conditions. The complex backgrounds are commonly encountered for high-resolution (HR) SAR ship detection data set, and they greatly influence the detection performance of ships. In recent years, deep neural networks (DNNs) have made substantial improvements on detection by adopting data augmentation. However, the improvement is limited since the models are sensitive to noise. To address this problem, a Fast Saliency-based Filtering algorithm (FSF) is proposed to filter out interference information. The FSF method is inspired by the filtering mechanisms of the human brain, which help people filter out target-irrelevant information fast to better extract target-relevant information. The FSF includes two parts of the bottom-up process and the top-down process. The bottom-up process is used to extract a saliency map of an input image, and the other one is used to filter out target-irrelevant information based on the saliency map. The FSF can be a front-end preprocessing module of DNNs to fast filter out target-irrelevant information and obtain a primary priority map of an input image. Experimental results demonstrate that our brain-inspired FSF method obtains obvious improvement of detection performance on AIR-SARShip-1.0.