Saved in:
Bibliographic Details
Main Authors: Wang, Zhifeng, Zhang, Kaihao, Sankaranarayana, Ramesh
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2408.04235
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911982083375104
author Wang, Zhifeng
Zhang, Kaihao
Sankaranarayana, Ramesh
author_facet Wang, Zhifeng
Zhang, Kaihao
Sankaranarayana, Ramesh
contents This paper introduces LLDif, a novel diffusion-based facial expression recognition (FER) framework tailored for extremely low-light (LL) environments. Images captured under such conditions often suffer from low brightness and significantly reduced contrast, presenting challenges to conventional methods. These challenges include poor image quality that can significantly reduce the accuracy of emotion recognition. LLDif addresses these issues with a novel two-stage training process that combines a Label-aware CLIP (LA-CLIP), an embedding prior network (PNET), and a transformer-based network adept at handling the noise of low-light images. The first stage involves LA-CLIP generating a joint embedding prior distribution (EPD) to guide the LLformer in label recovery. In the second stage, the diffusion model (DM) refines the EPD inference, ultilising the compactness of EPD for precise predictions. Experimental evaluations on various LL-FER datasets have shown that LLDif achieves competitive performance, underscoring its potential to enhance FER applications in challenging lighting conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2408_04235
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LLDif: Diffusion Models for Low-light Emotion Recognition
Wang, Zhifeng
Zhang, Kaihao
Sankaranarayana, Ramesh
Computer Vision and Pattern Recognition
This paper introduces LLDif, a novel diffusion-based facial expression recognition (FER) framework tailored for extremely low-light (LL) environments. Images captured under such conditions often suffer from low brightness and significantly reduced contrast, presenting challenges to conventional methods. These challenges include poor image quality that can significantly reduce the accuracy of emotion recognition. LLDif addresses these issues with a novel two-stage training process that combines a Label-aware CLIP (LA-CLIP), an embedding prior network (PNET), and a transformer-based network adept at handling the noise of low-light images. The first stage involves LA-CLIP generating a joint embedding prior distribution (EPD) to guide the LLformer in label recovery. In the second stage, the diffusion model (DM) refines the EPD inference, ultilising the compactness of EPD for precise predictions. Experimental evaluations on various LL-FER datasets have shown that LLDif achieves competitive performance, underscoring its potential to enhance FER applications in challenging lighting conditions.
title LLDif: Diffusion Models for Low-light Emotion Recognition
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2408.04235