Unsupervised Domain Adaptation for Learning Eye Gaze from a Million Synthetic Images: An Adversarial Approach
Explore this paper's citation graph
Summary
This paper initially learns a gaze estimator on annotated synthetic samples rendered from a 3D game engine and then adapt the features of unannotated real samples via a zero-sum minmax adversarial game against a domain discriminator following the recent paradigm of generative adversarial networks.
- Type
- preprint
- Published
- 2018-10-18
- Cited by
- 6
- References
- 45
- Access
- Open access
- OpenAlex
- https://openalex.org/W2897148684
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:53017696
Keywords
Computer science, Artificial intelligence, Machine learning, Domain (mathematical analysis), Benchmark (surveying)
References
- Adaptation
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Deep Domain Confusion: Maximizing for Domain Invariance
- Beyond the shortest path: Unsupervised domain adaptation by Sampling Subspaces along the Spline Flow
- Action recognition with trajectory-pooled deep-convolutional descriptors
- Gaze locking: passive eye contact detection for human-object interaction
- Learning-by-Synthesis for Appearance-Based 3D Gaze Estimation
- Adaptive Linear Regression for Appearance-Based Gaze Estimation
- Appearance-based gaze estimation in the wild
- Manifold Alignment for Person Independent Appearance-Based Gaze Estimation
- Dropout: a simple way to prevent neural networks from overfitting
- ImageNet Large Scale Visual Recognition Challenge
- Domain adaptation for object recognition: An unsupervised approach
- ImageNet classification with deep convolutional neural networks
- Return of Frustratingly Easy Domain Adaptation
- Deep Residual Learning for Image Recognition
- A Kernel Two-Sample Test
- Deep multi-scale video prediction beyond mean square error
- Learning a Deep Model for Human Action Recognition from Novel Viewpoints
- Learning an appearance-based gaze estimator from one million synthesised images
Cited by
- Realtime and Accurate 3D Eye Gaze Capture with DCNN-Based Iris and Pupil Segmentation
- WormPose: Image synthesis and convolutional networks for pose estimation in C. elegans
- Synthetic-to-Real Domain Adaptation for Lane Detection
- Gaze from Origin: Learning for Generalized Gaze Estimation by Embedding the Gaze Frontalization Process
- SynthEthics: Ensuring Digital Ethics and Performance with a Design Theory for Using Synthetic Image Data in Digital Health Deep Learning
- From Pixels to Generalization: Ensuring Information Security and Model Performance with Design Principles for Synthetic Image Data in Deep Learning
Related papers
- Learning from Simulated and Unsupervised Images through Adversarial Training
- Large Scale Unsupervised Domain Adaptation of Segmentation Networks with Adversarial Learning
- Improving the realism of synthetic images through a combination of adversarial and perceptual losses
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks
- Unsupervised Domain Adaptation: from Simulation Engine to the RealWorld