Scholar
Görkay Aydemir
Google Scholar ID: AtT2D54AAAAJ
Koc University
Citations & Impact
Citations
175
H-index
6
i10-index
2
Publications
8
Co-authors
2
list available
Publications
Resume (English only)
Academic Achievements
- Track-On2: Enhancing Online Point Tracking with Memory (under submission): Introduces Track-On2 with architectural refinements and improved memory mechanisms, achieving new SOTA in online point tracking with higher FPS and lower memory usage
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025): Presents a simple transformer-based model for causal long-term point tracking using spatial and context memory, setting new SOTA across multiple benchmarks among both online and offline methods
- Robust Bird’s Eye View Segmentation by Adapting DINOv2 (VCAD Workshop ECCV 2024): Enhances BEV perception robustness in autonomous driving by adapting DINOv2 with LoRA to handle brightness changes, adverse weather, and camera failures
- Can Visual Foundation Models Achieve Long-term Point Tracking? (EVAL-FoMo Workshop ECCV 2024): Evaluates geometric awareness of VFM for point tracking; shows DINOv2 matches supervised models after light training
- Self-supervised Object-centric Learning for Videos (NeurIPS 2023): Proposes SOLV, the first fully unsupervised method for multi-object segmentation in real-world videos using object-centric learning
- ADAPT: Efficient Multi-Agent Trajectory Prediction with Adaptation (ICCV 2023): Introduces ADAPT with dynamic weight learning and adaptive head for efficient and accurate multi-agent trajectory forecasting
- Trajectory Forecasting on Temporal Graphs (Arxiv preprint 2022): Proposes a temporal graph representation for future agent location prediction in dynamic traffic scenes
- MSc Thesis: Online Long-term Point Tracking in the Foundation Model Era (Koç University, 2025): Studies online point tracking in the foundation model era; proposes Track-On which achieves SOTA among online methods and matches or surpasses offline trackers on seven benchmarks
