Estimated Posture for Human Elbow Flexion in Acoustic Environments

The present study proposes a novel approach to address the challenge of identifying smoking and phone usage behaviours among individuals amidst the intricate surroundings of construction sites. Specifically, the method involves the recognition of elbow flexor behaviours through posture estimation. The localization and detection of key points pertaining to the human upper body have been achieved through the application of AlphaPose. Subsequently, the present study introduces the HEFBD model, which is a mathematical framework that discriminates human elbow flexion behaviour. The model is constructed using human key points and is designed to identify the area that is relevant for recognising tiny objects while mitigating the impact on the complicated background. A method for image reconstruction with super-resolution is employed as a pre-processing technique for images that exhibit blurring. Furthermore, the performance of You Only Look Once (YOLO) has been enhanced through the incorporation of a tiny object identification layer and the integration of a convolutional inhibitory attention model, thereby improving its detection capabilities. The method has demonstrated an improvement in detection precision of 6.7
^%
and a reduction in false identification rates due to complex backgrounds of 12%. These outcomes outperform existing detect approaches and meet the requirements for immediate results.