VALSE关注的研究方向

l  Imaging and low-level related

n  Computational photography

n  Low-level and physics-based vision

l  3D related

n  3D from a single image and shape-from-x

n  Stereo, 3D from multiview and other sensors

n  SLAM 

n Computer Vision with 3D

l  Fundamental and Generic CV tasks

n  Representation learning

n  Detection and localization in 2D

n  Recognition and classification

n  Segmentation, grouping and shape

n  Motion and tracking

n  Scene text and document understanding

n  Scene analysis and understanding

n  Video analysis and understanding

n  Image and video retrieval

n  Image and video synthesis

n  Image and video manipulation detection and integrity method

n  Anomaly detection

l  Methods(esp. Learning) for Vision

n  Adversarial learning

n  Self-supervised Learning

n  Knowledge leveraging for CV

n  Transfer/Low-shot/Semi-supervised learning

n  Brain-inspired CV

n  Transformer/Attention for CV

n  Graph models for CV

l  New machine learning theory and methods

n  Meta-learning

n  Incremental learning/continual learning/curriculum learning

n  Machine learning architectures and formulations

n  Causality learning

n  Efficient training and inference methods

n  Optimization and learning methods

l  Reliable CV models

n  Explainable CV models

n  Fairness, accountability, transparency, and ethics in vision

n  Visual reasoning and logical representation

l  Human-centered CV

n  Face recognition related  

n  Biometrics (except face)

n  Vision-based affective computing (e.g., expression recognition)

n  Medical image analysis

n  Human re-ID

n  Action and behavior recognition

n  Gestures and body pose

l  Vision + X

n  Vision + language

n  Vision + audio

n  Vision + other modalities

l  Vision for X

n  X=机器人

n  X=自动驾驶汽车

n  X=无人机

n  X=产品瑕疵检测

n  X=医疗(精神性疾病)

n  X=健康养老

n  X=Biology (animals…)

n  X=energy and power

l  Other topics

n  Heterogeneous pattern recognition

n  Occlusion-robust CV models