About this document
Visual Prompting for Emotion Recognition by shadril.shifat is a document available to read on EtoBox.
The document presents a novel Set-of-Vision (SoV) prompting approach to enhance emotion recognition in Vision Large Language Models (VLLMs) by incorporating spatial information such as bounding boxes and facial landmarks. This method improves the accuracy of detecting and categorizing emotions in images, addressing limitations of traditional visual prompting techniques that often overlook spatial context. Experimental results demonstrate that SoV prompts significantly enhance the performance of VLLMs in rec
- Author
- shadril.shifat
- Language
- EN