Exploring Data Aggregation and Transformations to Generalize across Visual Domains

Abstract

Computer vision has flourished in recent years thanks to Deep Learning advancements, fast and scalable hardware solutions and large availability of structured image data. Convolutional Neural Networks trained on supervised tasks with backpropagation learn to extract meaningful representations from raw pixels automatically, and surpass shallow methods in image understanding. Though convenient, data-driven feature learning is prone to dataset bias: a network learns its parameters from training signals alone, and will usually perform poorly if train and test distribution differ. To alleviate this problem, research on Domain Generalization (DG), Domain Adaptation (DA) and their variations is increasing. This thesis contributes to these research topics by presenting novel and effective ways to solve the dataset bias problem in its various settings. We propose new frameworks for Domain Generalization and Domain Adaptation which make use of feature aggregation strategies and visual transformations via data-augmentation and multi-task integration of self-supervision. We also design an algorithm that adapts an object detection model to any out of distribution sample at test time. With through experimentation, we show how our proposed solutions outperform competitive state-of-the-art approaches in established DG and DA benchmarks.

arXiv commentPhD thesis

References in corpus (15)

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• Learning Transferable Features with Deep Adaptation Networks
• Semantic Foggy Scene Understanding with Synthetic Data
• Unsupervised Domain Adaptation with Residual Transfer Networks
• A Survey on Multi-Task Learning
• Domain Generalization via Invariant Feature Representation
• Adversarial Discriminative Domain Adaptation
• A Collection of Definitions of Intelligence
• Self-supervised Domain Adaptation for Computer Vision Tasks
• Trace Norm Regularised Deep Multi-Task Learning
• Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection
• Self-Training and Adversarial Background Regularization for Unsupervised Domain Adaptive One-Stage Object Detection
• Episodic Training for Domain Generalization
• TWINs: Two Weighted Inconsistency-reduced Networks for Partial Domain Adaptation
• Deep Domain-Adversarial Image Generation for Domain Generalisation