The recent developments in the research area of semi-supervised and domain adaptation techniques for various applications, such as semantic segmentation, object detection, and network intrusion detection, have shown significant advancements. The field is moving towards more efficient and practical solutions that leverage both labeled and unlabeled data to reduce the dependency on extensive manual annotations. Innovations in feature augmentation, contrastive learning, and knowledge transfer are enhancing model robustness and performance across different domains. Notably, the integration of Vision Transformers with domain adaptation strategies is proving to be a powerful approach for improving model generalization. Additionally, the development of interactive and feedback-driven platforms for model refinement is demonstrating substantial gains in annotation efficiency without compromising quality. These trends indicate a shift towards more adaptable and user-friendly tools that can be applied across diverse applications, from medical imaging to autonomous driving.
Efficient Semi-Supervised and Domain Adaptation Techniques
Sources
The Last Mile to Supervised Performance: Semi-Supervised Domain Adaptation for Semantic Segmentation
GloFinder: AI-empowered QuPath Plugin for WSI-level Glomerular Detection, Visualization, and Curation
Improving Batch Normalization with TTA for Robust Object Detection in Self-Driving
Co-Learning: Towards Semi-Supervised Object Detection with Road-side Cameras
Knowledge-Data Fusion Based Source-Free Semi-Supervised Domain Adaptation for Seizure Subtype Classification
Feedback-driven object detection and iterative model improvement
SOUL: A Semi-supervised Open-world continUal Learning method for Network Intrusion Detection
Global Average Feature Augmentation for Robust Semantic Segmentation with Transformers
LoCo: Low-Contrast-Enhanced Contrastive Learning for Semi-Supervised Endoscopic Image Segmentation
Semi-Supervised Transfer Boosting (SS-TrBoosting)
Expanding Deep Learning-based Sensing Systems with Multi-Source Knowledge Transfer
TransAdapter: Vision Transformer for Feature-Centric Unsupervised Domain Adaptation
MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detection
Reflective Teacher: Semi-Supervised Multimodal 3D Object Detection in Bird's-Eye-View via Uncertainty Measure