KEYWORDS: Target detection, Deformation, Convolution, Animals, Animal model studies, Motion detection, Motion models, Visual process modeling, Detection and tracking algorithms, Education and training
For motion target detection in dynamic backgrounds, most traditional methods have drawbacks, such as long computation time or high limitation on the background. This paper proposes a deformable convolution-based motion target detection algorithm by replacing part of the C3 module in the YOLOv5 feature extraction layer with deformable convolution (DCNv3) to introduce long distance dependence and adaptive spatial aggregation, and adding an ECA attention mechanism to reduce the effect of background variation. The the accuracy, recall and mAP of the improved YOLOv5s algorithm increases by 1.6%, 3.8% and 2.3% respectively, and is able to identify moving targets in dynamic backgrounds more accurately than the original algorithm.
The relatively large variation in samples from different categories in the object detection dataset leads to the problem of sample imbalance, which in turn affects the accuracy of the object detector. In this paper, a dynamically weighted approach to label assignment is proposed to enable model to suppress abnormal samples and mine hard samples during training, thus solving the problem of sample imbalance. Different dynamic weighting strategies are applied to positive and negative samples to calculate the classification loss, and a dynamically weighted EIOU method is used to calculate the localization loss. In addition, modifying the single convolutional prediction head to a dual prediction head structure enables the model to focus more on performing classification and regression tasks. This structure uses a fully-connected head to predict the classification results and a convolutional head to predict the localisation results. Experiments conducted on PASCAL VOC2007+12 show that the proposed method can help YOLOv5 to achieve better accuracy and reach 72.3 mAP in 100 epochs, an improvement of 2.4 mAP over the original model. Furthermore, we have done several experiments on MSCOCO to demonstrate the effectiveness of the proposed dynamic weighting method, which has implications for building faster and stronger object detectors.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.