Synthetic data provides a scalable alternative to real-world datasets for training face recognition models, particularly under challenging conditions such as low resolution, occlusion, and masks. Yet, most approaches lack diversity and fail to generalize effectively. We propose Ensemble Feedback Controllable Synthesis...
Saransh Suri, Shivang Agarwal, M. Vatsa et al.· 0 citations
CricRAG, a retrieval-augmented framework that aligns VLMs with skill-appropriate benchmarks for personalized cricket coaching, is presented, finding that by retrieving similar-but-better techniques as reference points, it can guide VLMs to provide developmentally appropriate feedback that mirrors human coaching practic...
IAB edited achieves state-of-the-art instruction adherence performance on RealEdit and EMU Edit benchmarks based on embedding-based metrics and shows that gradient-aligned VLM distillation holds up under real-world-like surveillance and occlusion conditions.
Chiranjeev Chiranjeev, Muskan Dosi, M. Vatsa et al.· 0 citations
The Confidence-Weighted Numeric Precision Score (\cwnps), which aggregates three calibrated detectors and discounts uncertain proposals, is proposed for scalable evaluation and developed a process model in which requested instances compete for a finite set of resolvable image regions.
Sandeep Wadhwa, M. Vatsa, Richa Singh et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
