YoHa - A practical hand tracking engine.

Last update: Jan 06, 2023

Overview

YoHa

A practical hand tracking engine.

Quick Links:

Installation

npm install @handtracking.io/yoha

Please note:

You need to serve the files from node_modules/@handtracking.io/yoha. (Webpack Example)
You need to serve your page with https. (Webpack Example)
You should use cross-origin isolation as it improves the engine's performance in certain scenarios. (Webpack Example)

Description

YoHa is a hand tracking engine that is built with the goal of being a versatile solution in practical scenarios where hand tracking is employed to add value to an application. While ultimately the goal is to be a general purpose hand tracking engine supporting any hand pose, the engine evolves around specific hand poses that users/developers find useful. These poses are detected by the engine which allows to build applications with meaningful interactions. See the demo for an example.

YoHa is currently only available for the web.

YoHa is currently in beta.

About the name: YoHa is short for ("Your Hand Tracking").

Technical Details

YoHa was built from scratch. It uses a custom neural network trained using a custom dataset. The backbone for the inference in the browser is currently TensorFlow.js

Features:

Detection of 21 2D-landmark coordinates (single hand).
Hand presence detection.
Hand orientation (left/right hand) detection.
Inbuilt pose detection.

Supported hand poses:

Pinch
Fist

Your desired pose is not on this list? Feel free to create a GitHub issue for it.

Performance

YoHa was built with performance in mind. It is able to provide realtime user experience on a broad range of laptops and desktop devices. The performance on mobile devices is not great which hopefuly will change with the further development of inference frameworks like TensorFlow.js

Please note that native inference speed can not be compared with the web inference speed. Differently put, if you were to run YoHa natively it would be much faster than via the web browser.

Alternatives

The most prominent hand tracking solution for the web is from mediapipe. It is a very general and performant solution that keeps evolving and supports a lot of different deployment methods. In terms of performance the solution from mediapipe is faster. There is reason to believe that this is due to mediapipe using a different and closed source inference engine.

About

Hey, I'm Benjamin. I started out making this project because I wanted to make the web more interactive (especially due to covid imposed home office). Existing solutions did not offer what I was looking for so I built my own.

[email protected]

Comments

[Bug] Incorrect statement in README

Regarding note on MediaPipe:

There is reason to believe that this is due to mediapipe using a different and closed source inference engine.

It doesn't
MediaPipe HandDetect and HandLandmarks are completely standard TF models
(I'm using them without issues, just converted latest version of HandLandmarks sparse-variation from TFLite to TFJS Graph Model)

What is less open is their post-processing which is done using custom WASM backend so it
accelerates things like calculating L2Norm, buffering and smoothing of outputs, etc.
But if their model is used with WebGL backend, all that is a no-op

MediaPipe does have some more proprietary models such as their Holistic one as it uses custom TF ops to
link multiple models into single pipeline using attention for details processing
(If you're interested, custom ops are https://github.com/PINTO0309/PINTO_model_zoo/issues/143#issuecomment-938572434)

And even all that is not closed source, its just implemented in C (sources available) and compiled to WASM, for example https://github.com/google/mediapipe/blob/master/mediapipe/util/tflite/operations/transform_landmarks.cc

Anyhow, great job but I do wish you'd open source actual implementation
bug

opened by vladmandic 5
Custom model training

Big thanks for making your project open source with MIT license.

Are you planning to release details about the training of the model? Didn't see them (sorry if I missed them) but I think it would be great if you could share briefly the steps required to create and train the model with a custom data set, in order to also create gestures.

Thank you

opened by Lykos94 2
local demo question

I want to run the demo locally,and I followed your suggestions.then the question happended:

(base)MacBook-Pro ~ % git clone https://github.com/handtracking-io/yoha &&
cd yoha &&
yarn &&
yarn start

Cloning into 'yoha'... remote: Enumerating objects: 303, done. remote: Counting objects: 100% (303/303), done. remote: Compressing objects: 100% (180/180), done. remote: Total 303 (delta 129), reused 273 (delta 109), pack-reused 0 Receiving objects: 100% (303/303), 13.28 MiB | 78.00 KiB/s, done. Resolving deltas: 100% (129/129), done. zsh: command not found: yarn
bug

opened by Trueman1997 1
Test the demo locally

First of all, I'm a big fan of your project. I want to try and understand it even thought I lack most basic knowledge regarding HTML, JavaScript and TypeScript.

Using VS Code, I opened demo\draw\index.htlm with Live Server. The result on my browser don't show any picture nor camera, only text.

Could you please give me or redirect me to a brief tutorial on how to test your demo locally ?

Thanks you in advance

opened by Uspectacle 1
Difficulties detecting (or ignoring!) middle finger pinching

When playing around with the demo, I observed that it does really well for detecting index finger pinching, but struggles with middle finger pinching - to the point where I'm not even sure if it's intended to draw with middle finger pinching or not!

The video is a nice demo of the issue, but just generally if you move your hand while playing with middle-finger pinch I saw a lot of oddities and inconsistencies.

https://user-images.githubusercontent.com/1050652/137144907-9f52cbe0-f417-464b-a048-79fba116c8bf.mp4

opened by aidanhs 1
Bump minimist from 1.2.5 to 1.2.6
Bumps minimist from 1.2.5 to 1.2.6.

Commits

7efb22a 1.2.6

ef88b93 security notice for additional prototype pollution issue

c2b9819 isConstructorOrProto adapted from PR

bc8ecee test from prototype pollution PR

See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
Bump node-forge from 1.2.1 to 1.3.1 in /example
Bumps node-forge from 1.2.1 to 1.3.1.

Changelog

Sourced from node-forge's changelog.

1.3.1 - 2022-03-29

Fixes

RFC 3447 and RFC 8017 allow for optional DigestAlgorithm NULL parameters for sha* algorithms and require NULL paramters for md2 and md5 algorithms.

1.3.0 - 2022-03-17

Security

Three RSA PKCS#1 v1.5 signature verification issues were reported by Moosa Yahyazadeh ([email protected]).

HIGH: Leniency in checking digestAlgorithm structure can lead to signature forgery.

The code is lenient in checking the digest algorithm structure. This can allow a crafted structure that steals padding bytes and uses unchecked portion of the PKCS#1 encoded message to forge a signature when a low public exponent is being used. For more information, please see "Bleichenbacher's RSA signature forgery based on implementation error" by Hal Finney.

CVE ID: CVE-2022-24771

GHSA ID: GHSA-cfm4-qjh2-4765

HIGH: Failing to check tailing garbage bytes can lead to signature forgery.

The code does not check for tailing garbage bytes after decoding a DigestInfo ASN.1 structure. This can allow padding bytes to be removed and garbage data added to forge a signature when a low public exponent is being used. For more information, please see "Bleichenbacher's RSA signature forgery based on implementation error" by Hal Finney.

CVE ID: CVE-2022-24772

GHSA ID: GHSA-x4jg-mjrx-434g

MEDIUM: Leniency in checking type octet.

DigestInfo is not properly checked for proper ASN.1 structure. This can lead to successful verification with signatures that contain invalid structures but a valid digest.

CVE ID: CVE-2022-24773

GHSA ID: GHSA-2r2c-g63r-vccr

Fixed

[asn1] Add fallback to pretty print invalid UTF8 data.

[asn1] fromDer is now more strict and will default to ensuring all input bytes are parsed or throw an error. A new option parseAllBytes can disable this behavior.

NOTE: The previous behavior is being changed since it can lead to security issues with crafted inputs. It is possible that code doing custom DER parsing may need to adapt to this new behavior and optional flag.

[rsa] Add and use a validator to check for proper structure of parsed ASN.1

... (truncated)

Commits

a0a4a42 Release 1.3.1.

a33830f Update changelog.

740954d Allow optional DigestAlgorithm parameters.

56f4316 Allow DigestInfo.DigestAlgorith.parameters to be optional

cbf0bd5 Start 1.3.1-0.

6c5b901 Release 1.3.0.

0f3972a Update changelog.

dc77b39 Fix error checking.

bb822c0 Add advisory links.

d4395fe Update changelog.

Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
Bump minimist from 1.2.5 to 1.2.6 in /example
⚠️ Dependabot is rebasing this PR ⚠️

Rebasing might not happen immediately, so don't worry if this takes some time.

Note: if you make any changes to this PR yourself, they will take precedence over the rebase.

Bumps minimist from 1.2.5 to 1.2.6.

Commits

7efb22a 1.2.6

ef88b93 security notice for additional prototype pollution issue

c2b9819 isConstructorOrProto adapted from PR

bc8ecee test from prototype pollution PR

See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
Bump node-forge from 1.2.1 to 1.3.1
Bumps node-forge from 1.2.1 to 1.3.1.

Changelog

Sourced from node-forge's changelog.

1.3.1 - 2022-03-29

Fixes

RFC 3447 and RFC 8017 allow for optional DigestAlgorithm NULL parameters for sha* algorithms and require NULL paramters for md2 and md5 algorithms.

1.3.0 - 2022-03-17

Security

Three RSA PKCS#1 v1.5 signature verification issues were reported by Moosa Yahyazadeh ([email protected]).

HIGH: Leniency in checking digestAlgorithm structure can lead to signature forgery.

The code is lenient in checking the digest algorithm structure. This can allow a crafted structure that steals padding bytes and uses unchecked portion of the PKCS#1 encoded message to forge a signature when a low public exponent is being used. For more information, please see "Bleichenbacher's RSA signature forgery based on implementation error" by Hal Finney.

CVE ID: CVE-2022-24771

GHSA ID: GHSA-cfm4-qjh2-4765

HIGH: Failing to check tailing garbage bytes can lead to signature forgery.

The code does not check for tailing garbage bytes after decoding a DigestInfo ASN.1 structure. This can allow padding bytes to be removed and garbage data added to forge a signature when a low public exponent is being used. For more information, please see "Bleichenbacher's RSA signature forgery based on implementation error" by Hal Finney.

CVE ID: CVE-2022-24772

GHSA ID: GHSA-x4jg-mjrx-434g

MEDIUM: Leniency in checking type octet.

DigestInfo is not properly checked for proper ASN.1 structure. This can lead to successful verification with signatures that contain invalid structures but a valid digest.

CVE ID: CVE-2022-24773

GHSA ID: GHSA-2r2c-g63r-vccr

Fixed

[asn1] Add fallback to pretty print invalid UTF8 data.

[asn1] fromDer is now more strict and will default to ensuring all input bytes are parsed or throw an error. A new option parseAllBytes can disable this behavior.

NOTE: The previous behavior is being changed since it can lead to security issues with crafted inputs. It is possible that code doing custom DER parsing may need to adapt to this new behavior and optional flag.

[rsa] Add and use a validator to check for proper structure of parsed ASN.1

... (truncated)

Commits

a0a4a42 Release 1.3.1.

a33830f Update changelog.

740954d Allow optional DigestAlgorithm parameters.

56f4316 Allow DigestInfo.DigestAlgorith.parameters to be optional

cbf0bd5 Start 1.3.1-0.

6c5b901 Release 1.3.0.

0f3972a Update changelog.

dc77b39 Fix error checking.

bb822c0 Add advisory links.

d4395fe Update changelog.

Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
Bump follow-redirects from 1.14.7 to 1.14.8 in /example
Bumps follow-redirects from 1.14.7 to 1.14.8.

Commits

3d81dc3 Release version 1.14.8 of the npm package.

62e546a Drop confidential headers across schemes.

See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
Bump follow-redirects from 1.14.7 to 1.14.8
Bumps follow-redirects from 1.14.7 to 1.14.8.

Commits

3d81dc3 Release version 1.14.8 of the npm package.

62e546a Drop confidential headers across schemes.

See full diff in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

@dependabot rebase will rebase this PR

@dependabot recreate will recreate this PR, overwriting any edits that have been made to it

@dependabot merge will merge this PR after your CI passes on it

@dependabot squash and merge will squash and merge this PR after your CI passes on it

@dependabot cancel merge will cancel a previously requested merge and block automerging

@dependabot reopen will reopen this PR if it is closed

@dependabot close will close this PR and stop Dependabot recreating it. You can achieve the same result by closing it manually

@dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)

@dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

@dependabot use these labels will set the current labels as the default for future PRs for this repo and language

@dependabot use these reviewers will set the current reviewers as the default for future PRs for this repo and language

@dependabot use these assignees will set the current assignees as the default for future PRs for this repo and language

@dependabot use this milestone will set the current milestone as the default for future PRs for this repo and language

You can disable automated security fix PRs for this repo from the Security Alerts page.

dependencies
opened by dependabot[bot] 0
z depth and multiple camera.

would like 2 camera's for zdepth or support for depth camera's , also it would be neat in the future possibly to have 4 camera's setup so that if a user turns around he can still use hands with vr headset.
enhancement

opened by netpipe 0
[Feature Request] Hand to mouse

I have recently become unable to use a computer because of RSI. I'm using Talon Voice and Vimium for typing and web-browsing, but mouse use via an Tobii 5 eye-tracker has proven inaccurate and difficult, head-tracking using e-viacam is a lot of head-waving across dual-monitors and gets achy quickly.

Handtracking could replace mouse-usage with pinch-to-click, perhaps pinch-hold for right-click and v-fingers up/down scroll without putting any pressure on the wrist.

Screen-coordinate calibration and camera placement can remove issues with the hand in the line-of-sight and be useful for people with access needs that make mouse-usage impossible.

Many people with RSI find ways for words via dictation (ie with Dragon) but mouse-input often requires holding something, which eventually aggravates the condition. I think this feature would help those with mouse-usage issues. (Also, thanks for putting this together!)
enhancement

opened by drawnograph 0
[Bug] Error in detecting pinch
"Draw" is triggered even when the person is not using the pinch gesture.

To Reproduce Show your palm(slightly slacked) and rotate it 90 degrees or more along the vertical axis.

Expected behavior Nothing should happen.

Screenshots If applicable, add screenshots to help explain your problem.

Desktop :

OS: Ubuntu

chrome ver 94.0.4606.81

Additional context When viewed through the camera the thumb and index finger appear to be close like a pinch and is being interpreted that way but in reality they are far apart in the other dimension(the axis into and out of screen). Just shaking my hand makes it draw.
bug
opened by herocharge 0

Releases(v1.1.0)

v1.1.0(Feb 24, 2022)
New backend

Added support for tfjs wasm as computational backend.

See here for a demo for each backend.

API changes

Some functions and arguments have been renamed to make the API more consistent.

Source code(tar.gz)
Source code(zip)
v1.0.0(Feb 4, 2022)
Open Source

The new API is now fully open source.

API Rework

The API has been reworked substantially to be simpler and less opinionated. For now Yoha only runs on the tfjs/webgl backend. tfjs/wasm will be added in a future-release.

Other

Tfjs is now not bundled within Yoha which should make working with Yoha and other TFJS based packages more convenient.

A simple usage example has been added here.

Source code(tar.gz)
Source code(zip)

Owner

GitHub Repository

An official PyTorch implementation of the TKDE paper "Self-Supervised Graph Representation Learning via Topology Transformations".

Self-Supervised Graph Representation Learning via Topology Transformations This repository is the official PyTorch implementation of the following pap

2 Oct 31, 2022

A very simple tool for situations where optimization with onnx-simplifier would exceed the Protocol Buffers upper file size limit of 2GB, or simply to separate onnx files to any size you want.

sne4onnx A very simple tool for situations where optimization with onnx-simplifier would exceed the Protocol Buffers upper file size limit of 2GB, or

10 Aug 30, 2022

[CVPR 2021] "Multimodal Motion Prediction with Stacked Transformers": official code implementation and project page.

mmTransformer Introduction This repo is official implementation for mmTransformer in pytorch. Currently, the core code of mmTransformer is implemented

232 Dec 31, 2022

PyTorch implementation of Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets

Simple PyTorch Implementation of "Grokking" Implementation of Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets Usage Running

15 Sep 29, 2022

Research code for Arxiv paper "Camera Motion Agnostic 3D Human Pose Estimation"

GMR(Camera Motion Agnostic 3D Human Pose Estimation) This repo provides the source code of our arXiv paper: Seong Hyun Kim, Sunwon Jeong, Sungbum Park

1 Feb 07, 2022

DiffQ performs differentiable quantization using pseudo quantization noise. It can automatically tune the number of bits used per weight or group of weights, in order to achieve a given trade-off between model size and accuracy.

Differentiable Model Compression via Pseudo Quantization Noise DiffQ performs differentiable quantization using pseudo quantization noise. It can auto

145 Dec 30, 2022

KAPAO is an efficient multi-person human pose estimation model that detects keypoints and poses as objects and fuses the detections to predict human poses.

KAPAO (Keypoints and Poses as Objects) KAPAO is an efficient single-stage multi-person human pose estimation model that models keypoints and poses as

664 Dec 30, 2022

This repo contains the source code and a benchmark for predicting user's utilities with Machine Learning techniques for Computational Persuasion

Machine Learning for Argument-Based Computational Persuasion This repo contains the source code and a benchmark for predicting user's utilities with M

4 Nov 07, 2022

LSTC: Boosting Atomic Action Detection with Long-Short-Term Context

LSTC: Boosting Atomic Action Detection with Long-Short-Term Context This Repository contains the code on AVA of our ACM MM 2021 paper: LSTC: Boosting

9 Oct 11, 2022

A general framework for inferring CNNs efficiently. Reduce the inference latency of MobileNet-V3 by 1.3x on an iPhone XS Max without sacrificing accuracy.

GFNet-Pytorch (NeurIPS 2020) This repo contains the official code and pre-trained models for the glance and focus network (GFNet). Glance and Focus: a

169 Oct 28, 2022

YoHa - A practical hand tracking engine.

Related tags

Overview

YoHa

A practical hand tracking engine.

Quick Links:

Installation

Description

Technical Details

Features:

Supported hand poses:

Performance

Alternatives

About

Comments

1.3.1 - 2022-03-29

Fixes

1.3.0 - 2022-03-17

Security

Fixed

1.3.1 - 2022-03-29

Fixes

1.3.0 - 2022-03-17

Security

Fixed

Releases(v1.1.0)

v1.1.0(Feb 24, 2022)

New backend

API changes

v1.0.0(Feb 4, 2022)

Open Source

API Rework

Other

Owner

An official PyTorch implementation of the TKDE paper "Self-Supervised Graph Representation Learning via Topology Transformations".

A very simple tool for situations where optimization with onnx-simplifier would exceed the Protocol Buffers upper file size limit of 2GB, or simply to separate onnx files to any size you want.

[CVPR 2021] "Multimodal Motion Prediction with Stacked Transformers": official code implementation and project page.

PyTorch implementation of Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets

Research code for Arxiv paper "Camera Motion Agnostic 3D Human Pose Estimation"

DiffQ performs differentiable quantization using pseudo quantization noise. It can automatically tune the number of bits used per weight or group of weights, in order to achieve a given trade-off between model size and accuracy.

KAPAO is an efficient multi-person human pose estimation model that detects keypoints and poses as objects and fuses the detections to predict human poses.

This repo contains the source code and a benchmark for predicting user's utilities with Machine Learning techniques for Computational Persuasion

LSTC: Boosting Atomic Action Detection with Long-Short-Term Context

A general framework for inferring CNNs efficiently. Reduce the inference latency of MobileNet-V3 by 1.3x on an iPhone XS Max without sacrificing accuracy.

A PyTorch implementation of EfficientDet.

The implementation of FOLD-R++ algorithm

Official Implementation for "StyleCLIP: Text-Driven Manipulation of StyleGAN Imagery" (ICCV 2021 Oral)

Code for the paper "Reinforcement Learning as One Big Sequence Modeling Problem"

PyTorch code for the ICCV'21 paper: "Always Be Dreaming: A New Approach for Class-Incremental Learning"

Rational Activation Functions - Replacing Padé Activation Units

End-to-End Referring Video Object Segmentation with Multimodal Transformers

PyTorch implementation for Graph Contrastive Learning with Augmentations

Semantic Image Synthesis with SPADE

Tilted Empirical Risk Minimization (ICLR '21)