用教师学生框架让暗光视频动作识别更准，单流推理也能达顶尖水平。
Seeing in the Dark: A Teacher-Student Framework for Dark Video Action Recognition via Knowledge Distillation and Contrastive Learning
- 教师用暗光与增强帧双流输入，动态融合关键时间片段。
- 学生仅用暗光帧，通过知识蒸馏获得多流性能，准确率超96%。
- 无需额外计算开销，适合实际部署的低光视频分析场景。
暗光视频中的动作识别因视觉退化而困难，关键时空细节易丢失。本文提出ActLumos教师-学生框架，在单流推理下实现多流级精度。教师模型采用原始暗光帧与Retinex增强帧双流输入，经共享权重的R(2+1)D-34主干网络处理，由动态特征融合（DFF）模块在每时刻自适应加权，突出最有效时序段；教师还引入监督对比损失（SupCon），强化类别边界。学生模型共享相同主干，仅使用暗光帧且测试时不进行融合。学生先在无标签暗光片段上进行两视图时空自监督预训练，再通过知识蒸馏从教师迁移多流知识。单流推理下，学生在ARID V1.0上达到96.92%（Top-1）、ARID V1.5上88.27%、Dark48上48.96%的最新准确率。消融实验表明：教师端的DFF优于单流或静态融合；知识蒸馏有效将性能转移至单流学生；两视图时空自监督优于仅空间或仅时序版本，且不增加推理成本。
原文摘要 · Abstract (English)
Action recognition in dark or low-light (under-exposed) videos is a challenging task due to visibility degradation, which can hinder critical spatiotemporal details. This paper proposes ActLumos, a teacher-student framework that attains single-stream inference while retaining multi-stream level accuracy. The teacher consumes dual stream inputs, which include original dark frames and retinex-enhanced frames, processed by weight-shared R(2+1)D-34 backbones and fused by a Dynamic Feature Fusion (DFF) module, which dynamically re-weights the two streams at each time step, emphasising the most informative temporal segments. The teacher is also included with a supervised contrastive loss (SupCon) that sharpens class margins. The student shares the R(2+1)D-34 backbone but uses only dark frames and no fusion at test time. The student is first pre-trained with self-supervision on dark clips of both datasets without their labels and then fine-tuned with knowledge distillation from the teacher, transferring the teacher's multi-stream knowledge into a single-stream model. Under single-stream inference, the distilled student attains state-of-the-art accuracy of 96.92% (Top-1) on ARID V1.0, 88.27% on ARID V1.5, and 48.96% on Dark48. Ablation studies further highlight the individual contributions of each component, i.e., DFF in the teacher outperforms single or static fusion, knowledge distillation (KD) transfers these gains to the single-stream student, and two-view spatio-temporal SSL surpasses spatial-only or temporal-only variants without increasing inference cost. The official website of this work is available at: https://github.com/HrishavBakulBarua/ActLumos
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。