Abstract
To prevent coal mine safety accidents caused by non-standard mining operations, higher requirements have been proposed for identifying miners' behavior in surveillance videos. In this study, a single-stage detection model, YOLOv8, is employed, which effectively predicts the center of objects and is more suitable for recognizing miner behavior in complex environments. First, the video is segmented sampled to capture long-term information. Afterwards, deformable convolutions are integrated, applying offset learning to enhance fine-grained feature extraction and simultaneously, the introduction of the parameter-free SimAM improves the saliency mapping of miner behavior. Furthermore, an improved cooperative loss function is used for high-quality prior box regression. To validate the performance of non-standard miner behavior recognition, the proposed approach is conducted on a self-built dataset and achieves a mAP of 95.7 %, accuracy of 95.3 %, and recall of 95.1 % compared to the state-of-the-arts. It maintains an effective balance between computational cost and recognition accuracy.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.