AutoVP: An Automated Visual Prompting Framework and Benchmark

arXiv:2310.08381v2 Announce Type: replace-cross
Abstract: Visual prompting (VP) is an emerging parameter-efficient fine-tuning approach to adapting pre-trained vision models to solve various downstream image-classification tasks. However, there has hitherto been little systematic study of the design space of VP and no clear benchmark for evaluating its performance. To bridge this gap, we propose AutoVP, an end-to-end expandable framework for automating VP design choices, along with 12 downstream image-classification tasks that can serve as a holistic VP-performance benchmark. Our design space …

arxiv automated benchmark cs.cv cs.lg framework prompting type visual visual prompting