基于姿态的变压器模型，提升孟加拉手语识别准确率与效率
BdSL-SPOTER: A Transformer-Based Framework for Bengali Sign Language Recognition with Cultural Adaptation
- 采用文化特化预处理与优化位置编码的四层变压器架构
- 在BdSLW60数据集上达97.92%准确率，较LSTM提升22.82%
- 参数少、计算量低，适合资源有限地区手语应用
我们提出BdSL-SPOTER，一种基于姿态的变压器框架，用于高精度、高效率识别孟加拉手语（BdSL）。该框架在SPOTER基础上引入文化特化预处理和紧凑的四层变压器编码器，采用可学习位置编码优化，并通过课程学习提升小样本下的泛化能力与收敛速度。在BdSLW60基准上，其验证准确率达97.92%，相比Bi-LSTM基线提升22.82%，同时保持低计算成本。参数更少、FLOPs更低、帧率更高，为真实场景无障碍应用提供可行方案，并可扩展至其他低资源区域性手语。
原文摘要 · Abstract (English)
We introduce BdSL-SPOTER, a pose-based transformer framework for accurate and efficient recognition of Bengali Sign Language (BdSL). BdSL-SPOTER extends the SPOTER paradigm with cultural specific preprocessing and a compact four-layer transformer encoder featuring optimized learnable positional encodings, while employing curriculum learning to enhance generalization on limited data and accelerate convergence. On the BdSLW60 benchmark, it achieves 97.92% Top-1 validation accuracy, representing a 22.82% improvement over the Bi-LSTM baseline, all while keeping computational costs low. With its reduced number of parameters, lower FLOPs, and higher FPS, BdSL-SPOTER provides a practical framework for real-world accessibility applications and serves as a scalable model for other low-resource regional sign languages.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。