TraveLER: A Multi-LMM Agent Framework for Video Question-Answering

arXiv:2404.01476v1 Announce Type: cross
Abstract: Recently, Large Multimodal Models (LMMs) have made significant progress in video question-answering using a frame-wise approach by leveraging large-scale, image-based pretraining in a zero-shot manner. While image-based methods for videos have shown impressive performance, a current limitation is that they often overlook how key timestamps are selected and cannot adjust when incorrect timestamps are identified. Moreover, they are unable to extract details relevant to the question, instead providing general descriptions of the frame. To overcome …

abstract agent arxiv cs.ai cs.cl cs.cv cs.lg current framework image key large multimodal models lmm lmms multimodal multimodal models performance pretraining progress question scale type video videos wise zero-shot