An Underwater Single Target Tracking Method Using SiamRPN++ Based on Inverted Residual Bottleneck Block

Wang, Zongsheng; Wang, Jiaqi; Fan, Ruxin

doi:10.1109/access.2021.3056105

Cited by 12 publications

(5 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In addition to the above challenges, DL algorithms also face some other challenges in recognition and detection under some special circumstances, such as occlusion, illumination, shading, turbidity of the water and refraction of light. These environmental factors are rarely considered by the researchers, 197 and some of them cannot exist in the terrestrial environment, which may result in the decreased recognition accuracy (or more missing and false detections) under these special environments. In other words, DL solves most of the recognition and detection problems, but it is not a silver bullet that can solve all problems.…”

Section: Discussionmentioning

confidence: 99%

Deep learning for visual recognition and detection of aquatic animals: A review

Deng

et al. 2022

Reviews in Aquaculture

View full text Add to dashboard Cite

The ocean is an important ecosystem, and aquatic animals play an important role in the biological world, especially in aquaculture. How to accurately and intelligently recognise and detect aquatic animals is one of the urgent problems in the field of underwater biological detection. The wide applications of artificial intelligence (AI), especially deep learning (DL), provide new opportunities and challenges for the efficient and intelligent exploration of aquatic animals. DL has been widely used in the visual recognition and detection of terrestrial animals, but it is in the early stages of use for aquatic animals due to the complexity of underwater environment and the difficulty of data acquisition. Here, this article reviews the current application status of DL for aquatic animals, potential challenges and future directions. The key advances of DL algorithms applied to the visual recognition and detection of aquatic animals are generalised, including datasets, algorithms and performance. The applications of DL are summarised in aquatic animals, including image detection, video detection, species classification, biomass estimation, behaviour analysis and food safety. Furthermore, the challenges are summed up and classified in the object recognition and detection domain for aquatic animals. Finally, further research direction is discussed and the conclusions are drawn. The key advances of DL in the recognition and detection of aquatic animals will help to further excavate and extend the application of DL in the field of marine biological exploration.

show abstract

Section: Discussionmentioning

confidence: 99%

Deep learning for visual recognition and detection of aquatic animals: A review

Deng

et al. 2022

Reviews in Aquaculture

View full text Add to dashboard Cite

show abstract

“…Wang and colleagues artificially solved the problem of traditional tracking algorithms relying on manual feature extraction, and proposed an underwater single target tracking method using reverse residual bottleneck blocks. The results verified that the tracking speed of this method was improved to 73.74, and the accuracy was improved to 0.524 [15]. An N uses the conjoined network to detect and track visual objects, constructs the Siam network to classify moving objects, and uses the depth neural network and target detection to achieve multi-target tracking.…”

Section: Related Workmentioning

confidence: 66%

Deep Learning Based Target Tracking Algorithm Model for Athlete Training Trajectory

Wang

2022

Processes

View full text Add to dashboard Cite

The main function of the athlete tracking system is to collect the real-time competition data of the athletes. Deep learning is a research hotspot in the field of image and video. With the rapid development of science and technology, it has not only made a breakthrough in theory, but also achieved excellent results in practical application. SiamRPN (Siamese Region Proposal Network) is a single target tracking network model based on deep learning, which has high accuracy and fast operation speed. However, in long-term tracking, if the target is completely obscured and out of the sight of SiamRPN, the tracking of the network will be invalid. Considering the difficulty of long-term tracking, the algorithm is improved and tested by introducing channel attention mechanism and local global search strategy into SiamRPN. Experimental results show that this algorithm has higher accuracy and prediction average overlap rate than the original SiamRPN algorithm when performing tracking tasks on long-term tracking sequences. At the same time, the improved algorithm can still achieve good results in the case of target disappearance and other challenging factors. This study provides an important reference for the coaches of deep learning to realize long-term tracking of athletes.

show abstract

“…ROV can usually provide more energy and more powerful computing power, so it is feasible to use deep learning trackers on ROV to obtain more robust tracking performance. The reverse residual bottleneck block is added to SiamRPN++ to enhance the feature expression ability of the tracker to meet the challenge of underwater image degradation [34]. UStark [35] uses an image adaptive enhancement head to predict a set of enhancement parameters, and uses an enhancement module to process input images to improve the performance of the Stark tracker under different underwater image distortions.…”

Section: Underwater Object Trackingmentioning

confidence: 99%

Adjust Sample Imbalance and Exclude Similar Object in Underwater Object Tracking

Li¹,

Wang²,

Li³

et al. 2023

Preprint

View full text Add to dashboard Cite

Visual tracker includes network and post-processing. Despite the color distortion and low contrast of underwater images, advanced trackers can still be very competitive in underwater object tracking because deep learning empowers the networks to discriminate the appearance features of the target. However, underwater object tracking also faces another problem. Underwater targets such as fish and dolphins, usually appear in groups, and creatures of the same species usually have similar expressions of appearance features, so it is challenging to distinguish the weak differences characteristics only by the network itself. The existing detection-based post-processing only reflects the results of single frame detection, but cannot locate real targets among similar targets. In this paper, we propose a new post-processing strategy based on motion, which uses Kalman filter (KF) to maintain the motion information of the target and exclude similar targets around. Specifically, we use the KF predicted box and the candidate boxes in the response map and their confidence to calculate the candidate location score to find the real target. Our method does not change the network structure, nor does it perform additional training for the tracker. It can be quickly applied to other tracking fields with similar target problem. We improved SOTA trackers based on our method, and proved the effectiveness of our method on UOT100 and UTB180. The AUC of our method for OSTrack on similar subsequences is improved by more than 3% on average, and the precision and normalization precision are improved by more than 3.5% on average. It has been proved that our method has good compatibility in dealing with similar target problems and can enhance performance of the tracker together with other methods. More details can be found in: https://github.com/LiYunfengLYF/KF_in_underwater_trackers.

show abstract

An Underwater Single Target Tracking Method Using SiamRPN++ Based on Inverted Residual Bottleneck Block

Cited by 12 publications

References 34 publications

Deep learning for visual recognition and detection of aquatic animals: A review

Deep learning for visual recognition and detection of aquatic animals: A review

Deep Learning Based Target Tracking Algorithm Model for Athlete Training Trajectory

Adjust Sample Imbalance and Exclude Similar Object in Underwater Object Tracking

Contact Info

Product

Resources

About