Overview of Encord Index

Index is a data management system, that allows AI teams to visualize, sort, and control their data. The purpose is to give teams a better ability to control their curation process, enabling models to be put into production faster with better model quality.

When to use Encord Index?

Encord Index helps you visually understand and improve your data at all stages of your machine learning journey. Index unifies multimodal data from all local and cloud data sources for in-depth data management, visualization, search and granular curation.

Before going any further, you should know what a Collection is in Encord Index. Collections provide a way to save interesting groups of data units to support and guide your downstream workflow. For more information on Index Collections go here.

What data does Encord Index support?

Index supports analysis (Advanced Metrics and Embeddings) on images and videos up to 4K resolution. Performance is affected for images and videos over 4K.

• For optimal performance, we strongly recommend downscaling images and videos over 4K to 4K resolution.

• Videos import to Index at 1 FPS.

• You can specify specific frames you want imported from your videos.

• Contact us for more information on importing specific video frames to Index.

Features	Index
Image types	jpg, png, tiff
Video types	matroska and webm (vp8, vp9, libdav1d), mov, mp4, m4a, 3gp, 3g2, mj2 (libdav1d, h264, h265)
Audio types	mp3, .mp4, .wav, .m4a, .flac, .eac3
Number of images	10,000,000 per Folder (unlimited Folders)
Videos *	2 hours @ 30fps
Data exploration	✅
Similarity search	✅
Off-the-shelf quality metrics	✅
Image duplication detection	✅
Outlier detection	✅
Collections	✅
Integration with Encord Annotate	✅
Custom metadata	✅

• Contact Encord to discuss optimal strategies for indexing longer videos.

Data in Index

Overview

Standard Integrations

Data Registration

Data Access

Strict Client External Integration

Data Registration

Data Access

Encord Storage

Data Registration

Data Access