Abstract:
In recent years, 2D Human Pose Estimation (HPE) has become increasingly significant in complex computer vision tasks, encompassing the understanding of human behavior and...Show MoreMetadata
Abstract:
In recent years, 2D Human Pose Estimation (HPE) has become increasingly significant in complex computer vision tasks, encompassing the understanding of human behavior and interactions. Although faced with challenges like occlusion, unfavorable lighting, and motion blur, deep learning techniques have revolutionized 2D HPE by enabling automatic feature learning from data and improving generalization. However, prevalent 2D heatmap methods encounter issues like quantization errors, leading to complex post-processing needs. This paper proposed an innovative coordinated classification approach for 2D HPE, emphasizing enhanced prediction accuracy and model parameter optimization. Our novel model, AECA-PRNetCC, leverages a modified ResNet34 architecture, amplified by the Adaptive Efficient Channel Attention (AECA) mechanism, prioritizing essential features to enhance accuracy. Additionally, the model categorizes pixels into bins to mitigate quantization errors, thereby achieving superior accuracy. Comprehensive evaluations on the COCO dataset validate our model’s superior performance in both accuracy and computational efficiency compared to prevailing 2D HPE techniques.
Date of Conference: 29-30 November 2023
Date Added to IEEE Xplore: 12 December 2023
ISBN Information: