human Releases

A minor update adding Novita AI support and resolving specific spaCy language issues.

Jan 27, 2025

This release focuses on dependency updates and introduces new extraction techniques while fixing specific username extraction issues.

Dec 12, 2024

A feature-rich update introducing new AI providers (Upstage, Groq), AssemblyAI audio support, and significant improvements to document chunking and UI.

Dec 10, 2024

Sep 6, 2024

A maintenance release enhancing document support by adding .docx and GitLabReader, alongside improved embedding model support.

Jul 2, 2024

Refresh release focusing on toolkit upgrades including TypeScript 5.3.3, TFJS 4.17.0, and ESBuild 0.19.22. Includes custom CI platform refresh and new public config variables for face detector and iris scaling.

Feb 15, 2024

This release focuses on maintaining the project by repairing tests, updating dependencies, and improving documentation.

Oct 11, 2023

Major toolkit upgrade release updating to node v20, npm v10, TypeScript v5.1, and TFJS 4.11. Includes electron environment detection and several bug fixes.

Sep 18, 2023

Adds support for a new social media platform (memory.lol) to the extractor library.

Jul 24, 2023

Major release with significant optimizations including 4k resolution support, 30% smaller build size, and auto-backend selection. Introduces draw label templates and auto-brightness filter. Major refactoring of models namespace.

Nov 22, 2022

New webcam and video methods for continuous detection, redesigned segmentation with breaking changes, and new RVM body segmentation model.

Oct 13, 2022

Major update switching default backend to webgl, adding InsightFace model support, and introducing human.check() for kernel validation. No breaking changes despite being flagged as major release.

Sep 7, 2022

Optimized installation size and performance by switching to production dependencies and improving browser compatibility.

Jul 25, 2022

Updated face mesh models and introduced an attention-based model with enhanced performance and compatibility.

Jun 8, 2022

Introduces MediaPipe FaceMesh-with-Attention, a model pre-compile phase, 4K support, and new ElectronJS demos.

May 9, 2022

This release focuses on browser environment improvements, adding model caching using IndexedDB, fixing detection accuracy for occluded faces and hands, and enhancing Node.js compatibility for model loading.

Jan 20, 2022

Dec 15, 2021

Nov 8, 2021

Version 2.4.1 introduces time-based caching, an optional Anti-Spoof module, and utility class enhancements, alongside improved MoveNet post-processing and the removal of legacy models.

Oct 25, 2021

Oct 6, 2021