Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation

Related tags

Deep LearningPnP-GA
Overview

Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation

Python 3.6 Pytorch 1.5.0 CUDA 10.2 License CC BY-NC

Our paper is accepted by ICCV2021.

Teaser

Picture: Overview of the proposed Plug-and-Play (PnP) adaption framework for generalizing gaze estimation to a new domain.

Main image

Picture: The proposed architecture.


Results

Input Method DE→DM DE→DD DG→DM DG→DD
Face Baseline 8.767 8.578 7.662 8.977
Face Baseline + PnP-GA 5.529 ↓36.9% 5.867 ↓31.6% 6.176 ↓19.4% 7.922 ↓11.8%
Face ResNet50 8.017 8.310 8.328 7.549
Face ResNet50 + PnP-GA 6.000 ↓25.2% 6.172 ↓25.7% 5.739 ↓31.1% 7.042 ↓6.7%
Face SWCNN 10.939 24.941 10.021 13.473
Face SWCNN + PnP-GA 8.139 ↓25.6% 15.794 ↓36.7% 8.740 ↓12.8% 11.376 ↓15.6%
Face + Eye CA-Net -- -- 21.276 30.890
Face + Eye CA-Net + PnP-GA -- -- 17.597 ↓17.3% 16.999 ↓44.9%
Face + Eye Dilated-Net -- -- 16.683 18.996
Face + Eye Dilated-Net + PnP-GA -- -- 15.461 ↓7.3% 16.835 ↓11.4%

This repository contains the official PyTorch implementation of the following paper:

Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation
Yunfei Liu, Ruicong Liu, Haofei Wang, Feng Lu

Abstract: Deep neural networks have significantly improved appearance-based gaze estimation accuracy. However, it still suffers from unsatisfactory performance when generalizing the trained model to new domains, e.g., unseen environments or persons. In this paper, we propose a plugand-play gaze adaptation framework (PnP-GA), which is an ensemble of networks that learn collaboratively with the guidance of outliers. Since our proposed framework does not require ground-truth labels in the target domain, the existing gaze estimation networks can be directly plugged into PnP-GA and generalize the algorithms to new domains. We test PnP-GA on four gaze domain adaptation tasks, ETH-to-MPII, ETH-to-EyeDiap, Gaze360-to-MPII, and Gaze360-to-EyeDiap. The experimental results demonstrate that the PnP-GA framework achieves considerable performance improvements of 36.9%, 31.6%, 19.4%, and 11.8% over the baseline system. The proposed framework also outperforms the state-of-the-art domain adaptation approaches on gaze domain adaptation tasks.

Resources

Material related to our paper is available via the following links:

System requirements

  • Only Linux is tested, Windows is under test.
  • 64-bit Python 3.6 installation.

Playing with pre-trained networks and training

Config

You need to modify the config.yaml first, especially xxx/image, xxx/label, and xxx_pretrains params.

xxx/image represents the path of label file.

xxx/root represents the path of image file.

xxx_pretrains represents the path of pretrained models.

A example of label file is data folder. Each line in label file is conducted as:

p00/face/1.jpg 0.2558059438789034,-0.05467275933864655 -0.05843388117618364,0.46745964684693614 ... ...

Where our code reads image data form os.path.join(xxx/root, "p00/face/1.jpg") and reads ground-truth labels of gaze direction from the rest in label file.

Train

We provide three optional arguments, which are --oma2, --js and --sg. They repersent three different network components, which could be found in our paper.

--source and --target represent the datasets used as the source domain and the target domain. You can choose among eth, gaze360, mpii, edp.

--i represents the index of person which is used as the training set. You can set it as -1 for using all the person as the training set.

--pics represents the number of target domain samples for adaptation.

We also provide other arguments for adjusting the hyperparameters in our PnP-GA architecture, which could be found in our paper.

For example, you can run the code like:

python3 adapt.py --i 0 --pics 10 --savepath path/to/save --source eth --target mpii --gpu 0 --js --oma2 --sg

Test

--i, --savepath, --target are the same as training.

--p represents the index of person which is used as the training set in the adaptation process.

For example, you can run the code like:

python3 test.py --i -1 --p 0 --savepath path/to/save --target mpii

Citation

If you find this work or code is helpful in your research, please cite:

@inproceedings{liu2021PnP_GA,
  title={Generalizing Gaze Estimation with Outlier-guided Collaborative Adaptation},
  author={Liu, Yunfei and Liu, Ruicong and Wang, Haofei and Lu, Feng},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  year={2021}
}

Contact

If you have any questions, feel free to E-mail me via: lyunfei(at)buaa.edu.cn

Owner
Yunfei Liu
;-)
Yunfei Liu
PixelPyramids: Exact Inference Models from Lossless Image Pyramids (ICCV 2021)

PixelPyramids: Exact Inference Models from Lossless Image Pyramids This repository contains the PyTorch implementation of the paper PixelPyramids: Exa

Visual Inference Lab @TU Darmstadt 8 Dec 11, 2022
Object tracking using YOLO and a tracker(KCF, MOSSE, CSRT) in openCV

Object tracking using YOLO and a tracker(KCF, MOSSE, CSRT) in openCV File YOLOv3 weight can be downloaded

Ngoc Quyen Ngo 2 Mar 27, 2022
OpenAi's gym environment wrapper to vectorize them with Ray

Ray Vector Environment Wrapper You would like to use Ray to vectorize your environment but you don't want to use RLLib ? You came to the right place !

Pierre TASSEL 15 Nov 10, 2022
A fast and easy to use, moddable, Python based Minecraft server!

PyMine PyMine - The fastest, easiest to use, Python-based Minecraft Server! Features Note: This list is not always up to date, and doesn't contain all

PyMine 144 Dec 30, 2022
Autonomous racing with the Anki Overdrive

Anki Autonomous Racing Autonomous racing with the Anki Overdrive. Using the Overdrive-Python API (https://github.com/xerodotc/overdrive-python) develo

3 Dec 11, 2022
High-resolution networks and Segmentation Transformer for Semantic Segmentation

High-resolution networks and Segmentation Transformer for Semantic Segmentation Branches This is the implementation for HRNet + OCR. The PyTroch 1.1 v

HRNet 2.8k Jan 07, 2023
Generalized Data Weighting via Class-level Gradient Manipulation

Generalized Data Weighting via Class-level Gradient Manipulation This repository is the official implementation of Generalized Data Weighting via Clas

18 Nov 12, 2022
Code for the Population-Based Bandits Algorithm, presented at NeurIPS 2020.

Population-Based Bandits (PB2) Code for the Population-Based Bandits (PB2) Algorithm, from the paper Provably Efficient Online Hyperparameter Optimiza

Jack Parker-Holder 22 Nov 16, 2022
Fashion Entity Classification

Fashion-Entity-Classification - Fashion-MNIST is a dataset of Zalando's article images—consisting of a training set of 60,000 examples and a test set of 10,000 examples. Each example is a 28x28 grays

ADITYA SHAH 1 Jan 04, 2022
Trading environnement for RL agents, backtesting and training.

TradzQAI Trading environnement for RL agents, backtesting and training. Live session with coinbasepro-python is finaly arrived ! Available sessions: L

Tony Denion 164 Oct 30, 2022
Official code for "Towards An End-to-End Framework for Flow-Guided Video Inpainting" (CVPR2022)

E2FGVI (CVPR 2022) English | 简体中文 This repository contains the official implementation of the following paper: Towards An End-to-End Framework for Flo

Media Computing Group @ Nankai University 537 Jan 07, 2023
CTF challenges from redpwnCTF 2021

redpwnCTF 2021 Challenges This repository contains challenges from redpwnCTF 2021 in the rCDS format; challenge information is in the challenge.yaml f

redpwn 27 Dec 07, 2022
DCA - Official Python implementation of Delaunay Component Analysis algorithm

Delaunay Component Analysis (DCA) Official Python implementation of the Delaunay

Petra Poklukar 9 Sep 06, 2022
[ICCV 2021] Official Pytorch implementation for Discriminative Region-based Multi-Label Zero-Shot Learning SOTA results on NUS-WIDE and OpenImages

Discriminative Region-based Multi-Label Zero-Shot Learning (ICCV 2021) [arXiv][Project page coming soon] Sanath Narayan*, Akshita Gupta*, Salman Kh

Akshita Gupta 54 Nov 21, 2022
Rethinking Transformer-based Set Prediction for Object Detection

Rethinking Transformer-based Set Prediction for Object Detection Here are the code for the ICCV paper. The code is adapted from Detectron2 and AdelaiD

Zhiqing Sun 62 Dec 03, 2022
A Model for Natural Language Attack on Text Classification and Inference

TextFooler A Model for Natural Language Attack on Text Classification and Inference This is the source code for the paper: Jin, Di, et al. "Is BERT Re

Di Jin 418 Dec 16, 2022
DEMix Layers for Modular Language Modeling

DEMix This repository contains modeling utilities for "DEMix Layers: Disentangling Domains for Modular Language Modeling" (Gururangan et. al, 2021). T

Suchin 43 Nov 11, 2022
🏆 The 1st Place Submission to AICity Challenge 2021 Natural Language-Based Vehicle Retrieval Track (Alibaba-UTS submission)

AI City 2021: Connecting Language and Vision for Natural Language-Based Vehicle Retrieval 🏆 The 1st Place Submission to AICity Challenge 2021 Natural

82 Dec 29, 2022
Inverse Optimal Control Adapted to the Noise Characteristics of the Human Sensorimotor System

Inverse Optimal Control Adapted to the Noise Characteristics of the Human Sensorimotor System This repository contains code for the paper Schultheis,

2 Oct 28, 2022