Junlin Han
Scholar

Junlin Han

Google Scholar ID: 5L0Uj_IAAAAJ
Meta AI | University of Oxford
Computer visionMachine LearningArtificial Intelligence
Citations & Impact
All-time
Citations
1,043
 
H-index
14
 
i10-index
14
 
Publications
20
 
Co-authors
42
list available
Resume (English only)
Academic Achievements
  • NeurIPS 2025: 'Hallucination at a Glance' – a new dataset for fine-grained visual reasoning in MLLMs
  • SIGGRAPH 2025: '3D2EP' – 3D shape decomposition using parametric primitives
  • ICML 2025: 'Flex3D' – two-stage pipeline for high-quality 3D generation and reconstruction
  • CVPR 2025: paper on point-cloud segmentation
  • Best Paper Award in Remote Sensing (2024) for work on low-level vision
  • 3DV 2025: two papers – '3D-GPT' (large-scale 3D scene generation) and 'DreamBeast' (fantastical 3D animal generation)
  • ECCV 2024: 'VFusion3D' (scalable 3D generative/reconstruction models) and 'Unicorns'
  • ICCV 2023: 'Hyperbolic Audio-visual Zero-shot Learning'
  • ECCV 2022: 'Blind Image Decomposition (BID)' – a new low-level vision task
  • ICML 2022: 'You Only Cut Once (YOCO)' – data augmentation method
  • Top Reviewer at NeurIPS 2022
  • Co-organizing the 'Foundation Models in the Wild' workshops at ICLR 2025 and ICML 2024