Release Notes#

Current Release#

Version: 2025.2

This is the first release for multilevel video understanding. This microservice implements a novel multi-level framework designed specifically for video summarization, with specialized capabilities for processing long-form video content.

Features

Process video from local files or http(s) links.
Containerization with Docker.
RESTful API with FastAPI with support for concurrent requests.
Support specify video chunking method in user requests.
Support specify multi-level settings in user requests.
Support specify temporal enhancement settings in user requests.
Designed to work effortlessly with GenAI model servings that provide OpenAI-compatible APIs.

HW used for validation:

Intel® Xeon™ GOLD 6530
Intel® Arc™ Pro B60 Graphics * 4