Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window. Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window.

Batch Video Warehouse Overview

Batch Video Warehouse (also known as Video on Demand Warehouse) helps developers to:

Import video files from Cloud Storage.
Import video metadata.
Analyze video content embeddings.
Construct and deploy search indexes for AI-powered video search.

There are a few basic steps to use a Batch Video Warehouse:

Upload and analyze videos: Create a corpus, upload video assets, generate video embeddings by analyze the video assets. Additionally, developers can define data schema and annotations to enrich videos with supplementary metadata.
Build and deploy search indexes: Create an index and deploy the index to an index endpoint for search. A deployed index is managed by a dedicated pool of resources that scales with the size of the index.
Search, browse, and update index data: Perform semantic search for relevant video partitions on a deployed index with text queries, images, or filter criteria on the annotation metadata. Users can also browse the corpus or index contents through ListAssets and ViewIndexedAssets respectively. Batch Video Warehouse provides two ways for updating video assets in the index:
- Incremental update: Use the IndexAsset or RemoveIndexAsset APIs to update one video at a time. This method leads to limited throughput, but low update latency.
- Batch update: Use the UpdateIndex API with the asset_filter parameter to fully rebuild the index. This way is preferred when you have a large number of videos to add or remove from the index.

Supported video formats

Warehouse backends support popular formats such as MP4, AVI, FLV, MKV, MOV, MXF, WMV, and the formats decodable by FFmpeg.

The frontend, which is only for demo purposes, supports MP4 and MKV.

What's next

Start onboarding with the demo colab.