测试两种改进CorrNet的结构，发现效果不佳且冗余。
Do Transformer Temporal Heads and Post-Pooling Motion Gates Help CorrNet-based CSLR? An Empirical Study
- 用Transformer替换BiLSTM，性能未提升
- 设计轻量运动门（MotionGate），但失去选择性
- 提示结构改进需实证，不可盲目假设
CorrNet是连续手语识别（CSLR）的强基准模型，因其在视觉编码阶段建模帧间相关性。本文对复现的CorrNet系统进行两项自然扩展：用Transformer编码器替代BiLSTM时间头，以及在时间池化后注入运动线索。实验发现，即使调整训练策略，Transformer头也未能超越BiLSTM基线，且计算与运行成本几乎相同。针对第二项改进，设计了轻量级模块MotionGate，但实验显示其始终退化为恒等映射：门控失去运动选择性，注入的残差仅对池化特征产生微弱、非选择性扰动。结果表明，在基于相关性的编码之后显式注入运动信息基本冗余，提醒在CSLR中自然的架构扩展应经严格验证，而非默认有益。
原文摘要 · Abstract (English)
CorrNet is a strong baseline for continuous sign language recognition (CSLR) because it models inter-frame correlations inside the visual encoding stage. In this paper, we study two natural extensions of a reproduced CorrNet system: replacing the BiLSTM temporal head with a Transformer encoder, and injecting motion cues after temporal pooling. We find that the Transformer head does not outperform the BiLSTM baseline, even with a training strategy adjusted for the Transformer, and the two heads have almost the same computational and runtime cost. For the second extension, we design a lightweight module called MotionGate. In our experiments, MotionGate consistently collapses to an identity-like mapping: the gate loses motion selectivity, and the injected residual becomes a weak, non-selective perturbation of the pooled features. These results suggest that explicit motion injection after CorrNet's correlation-based encoding is largely redundant, and that natural-looking architectural extensions in CSLR should be tested carefully instead of being assumed to help.
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。