Meta-UDA: Unsupervised Domain Adaptive Thermal Object Detection using Meta-Learning

  title={Meta-UDA: Unsupervised Domain Adaptive Thermal Object Detection using Meta-Learning},
  author={V. S. Vibashan and Domenick Poster and Suya You and Shuowen Hu and Vishal M. Patel},
  journal={2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
Object detectors trained on large-scale RGB datasets are being extensively employed in real-world applications. However, these RGB-trained models suffer a performance drop under adverse illumination and lighting conditions. Infrared (IR) cameras are robust under such conditions and can be helpful in real-world applications. Though thermal cameras are widely used for military applications and increasingly for commercial applications, there is a lack of robust algorithms to robustly exploit the… 

Figures and Tables from this paper

Towards Online Domain Adaptive Object Detection

MemXformer is introduced - a cross-attention transformer-based memory module where items in the memory take advantage of domain shifts and record prototypical patterns of the target distribution, which enhances target-specific representation learning.

Few-shot Adaptive Object Detection with Cross-Domain CutMix

A data synthesis method that can solve the large domain gap problem by pasting a part of the target image onto the source image, and the position of the pasted region is aligned by utilizing the information of the object bounding box.

Open-Set Automatic Target Recognition

This work proposes an Open-set Automatic Target Recognition framework where it enables open-set recognition capability for ATR algorithms and introduces a plugin Category-aware Binary Classifier (CBC) module to effectively tackle unknown classes seen during inference.

Few-Shot Learning of Compact Models via Task-Specific Meta Distillation

A task-specific meta distillation that simultaneously learns two models in meta-learning: a large teacher model and a small student model that are jointly learned during meta-training.



Unsupervised Domain Adaption of Object Detectors: A Survey

This work describes in detail the domain adaptation problem for detection and presents an extensive survey of the various methods, and identifies multiple aspects of the problem that are most promising for future research.

Borrow From Anywhere: Pseudo Multi-Modal Object Detection in Thermal Imagery

This paper proposes a pseudo-multimodal object detector trained on natural image domain data to help improve the performance of object detection in thermal images and shows that the framework outperforms existing benchmarks without the explicit need for paired training examples from the two domains.

Adapting Object Detectors via Selective Cross-Domain Alignment

The key idea is to mine the discriminative regions, namely those that are directly pertinent to object detection, and focus on aligning them across both domains, and perform remarkably better than existing methods.

Domain Adaptive Faster R-CNN for Object Detection in the Wild

This work builds on the recent state-of-the-art Faster R-CNN model, and design two domain adaptation components, on image level and instance level, to reduce the domain discrepancy, based on $$-divergence theory.

Progressive Domain Adaptation for Object Detection

This paper proposes to bridge the domain gap with an intermediate domain and progressively solve easier adaptation subtasks by adopting adversarial learning to align distributions at the feature level.

Automatic Adaptation of Object Detectors to New Domains Using Self-Training

The usefulness of incorporating hard examples obtained from tracking, the advantage of using soft-labels via distillation loss versus hard-lab labels, and promising performance as a simple method for unsupervised domain adaptation of object detectors, with minimal dependence on hyper-parameters are demonstrated.

A Robust Learning Approach to Domain Adaptive Object Detection

A robust object detection framework that is resilient to noise in bounding box class labels, locations and size annotations is proposed that significantly improves the state-of-the-art on multiple domain adaptation scenarios on the SIM10K, Cityscapes and KITTI datasets.

SSTN: Self-Supervised Domain Adaptation Thermal Object Detection for Autonomous Driving

This work explores thermal object detection to model a view-invariant model representation by employing the self-supervised contrastive learning approach and proposes a deep neural network Self Supervised Thermal Network (SSTN) for learning the feature embedding to maximize the information between visible and infrared spectrum domain by contrastive learn.

TIRNet: Object detection in thermal infrared images for autonomous driving

A novel object detection approach, termed TIRNet, which is built upon convolutional neural network (CNN), which gets the state-of-the-art detection accuracy while maintains high detection efficiency.

Strong-Weak Distribution Alignment for Adaptive Object Detection

This work proposes an approach for unsupervised adaptation of object detectors from label-rich to label-poor domains which can significantly reduce annotation costs associated with detection, and designs the strong domain alignment model to only look at local receptive fields of the feature map.