Alibaba Cloud Big Data AI Platform

Nov 7, 2024 · Artificial Intelligence

How VideoCLIP‑XL Boosts Long‑Description Understanding in Video CLIP Models

VideoCLIP‑XL, a new video CLIP model introduced by Alibaba Cloud AI Platform and Sun Yat‑sen University, enhances long‑text description comprehension through a large‑scale VILD dataset, a text‑similarity guided principal component matching method, and novel DDR and HDR ranking tasks, achieving superior performance on multiple video‑text benchmarks.