egonet-method / README.md
WhynotGPT's picture
Update EgoNet fixed-camera instance Pretrain and SFT method gallery
5acbf5e verified
|
Raw History Blame Contribute Delete
1.63 kB
metadata
title: EgoNet · Foundation Hand
emoji: ✋
colorFrom: green
colorTo: purple
sdk: static
app_file: index.html
pinned: false

EgoNet / Foundation Hand method

Version: 2026-10-26 · fixed-camera instance Pretrain → egocentric SFT.

A static method gallery, including editable SVG, high-resolution PNG and PDF figures. This Space serves documentation only; it does not run model inference or expose training datasets, checkpoints, MANO assets, credentials or server files.

Maintained alongside mirrorbloom/EgoNet, under docs/site/. Source-repository access remains subject to its existing permissions. The diagram distinguishes the current instance Pretrain from subsequent SFT and separate camera-extrinsics experiments. It is not a live training status page.

Input illustrations are extracted RGB panels from Re:InterHand (Moon et al., NeurIPS Datasets and Benchmarks 2023), licensed CC BY-NC 4.0. See the official project and license. Changes: frame selection, panel extraction, resizing and diagram composition. The hand skeleton is a schematic illustration, not a prediction for the photographs. Neither the photos nor the illustration constitute new experimental results.

Photo sources: