FRU-Adapter: Frame Recalibration Unit Adapter for Dynamic Facial Expression Recognition
Dynamic facial expression recognition (DFER) is one of the most important challenges in computer vision, as it plays a crucial role in human–computer interaction. Re cently, adapter-based approaches have been introduced into DFER, and they have achieved remarkable success. However, the adapters still suffer from the following problems: over looking irrelevant frames and interference with pre-trained information. In this paper, we propose a frame recalibration unit adapter (FRU-Adapter) which combines the strengths of a frame recalibration unit (FRU) and temporal self-attention (T-SA) to address the afore mentioned issues. The FRU initially recalibrates the frames by emphasizing important frames and suppressing less relevant frames. The recalibrated frames are then fed into T-SA to capture the correlations between meaningful frames. As a result, the FRU-Adapter captures enhanced temporal dependencies by considering the irrelevant frames in a clip. Furthermore, we propose a method for attaching the FRU-Adapter to each encoder layer in parallel to reduce the loss of pre-trained information. Notably, the FRU-Adapter uses only 2% of the total training parameters per task while achieving an improved accuracy. Extended experiments on DFER tasks show that the proposed FRU-Adapter not only out performs the state-of-the-art models but also exhibits parameter efficiency. The source code will be made publicly available.
Code (1)
Tasks
Dynamic Facial Expression RecognitionFacial Expression RecognitionMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Perceive, Route and Modulate: Dynamic Pattern Recalibration for Time Series Forecasting
Local temporal patterns in real-world time series continuously shift, rendering globally shared transformations suboptimal. Current deep forecasting models, despite their scale and complexity, rely on fixed weight matric…
Time Series ForecastingFisheye-Calib-Adapter: An Easy Tool for Fisheye Camera Model Conversion
The increasing necessity for fisheye cameras in fields such as robotics and autonomous driving has led to the proposal of various fisheye camera models. While the evolution of camera models has facilitated the developmen…
Autonomous DrivingAdaptive Minds: Empowering Agents with LoRA-as-Tools
We investigate a framework in which LoRA adapters are treated as callable tools that a base language model can dynamically select and invoke. We hypothesize that, when adapters are trained to provide strong domain-specif…
AdaFuse: Accelerating Dynamic Adapter Inference via Token-Level Pre-Gating and Fused Kernel Optimization
The integration of dynamic, sparse structures like Mixture-of-Experts (MoE) with parameter-efficient adapters (e.g., LoRA) is a powerful technique for enhancing Large Language Models (LLMs). However, this architectural e…
S-LoRA: Serving Thousands of Concurrent LoRA Adapters
The "pretrain-then-finetune" paradigm is commonly adopted in the deployment of large language models. Low-Rank Adaptation (LoRA), a parameter-efficient fine-tuning method, is often employed to adapt a base model to a mul…
GPUparameter-efficient fine-tuning