Re:Cognize — Open-Set Comic Character Re-IdentificationA. Baranwal, M. Kataria, Y.S. Rawat, S. Vyas
Publications
Mostly benchmarks: careful tests of what vision-language models can and can't do.
PHOEBI: An Open-World Benchmark for Multi-Label Bacterial Identification in Phase-Contrast MicroscopyA. Baranwal, M.J. Hasan, S. Vyas
Vision-Language Models for Analog Gauge Reading: An Empirical Study of Specialization, Transfer and ReliabilityA. Mueez, A. Baranwal, J. Chaj-Mejia, G. Bhatia, J. Voelker, S. Vyas
BareBones: Benchmarking Zero-Shot Geometric Comprehension in VLMsA. Baranwal, V. Yadav, A. Rajora
ChemPro: A Progressive Chemistry Benchmark for Large Language ModelsA. Baranwal, S. Vyas
Robust Onion: Peeling Open Vocab Object Detectors Under NoiseP. Pathak, M. Karuppasamy, A. Baranwal, S. Vyas, Y.S. Rawat
ZeroBench: An Impossible Visual Benchmark for Contemporary Large Multimodal ModelsJ. Roberts, M.R. Taesiri, A. Sharma, A. Gupta, S. Roberts, I. Croitoru, A. Baranwal, et al.
Re:Verse — Can Your VLM Read a Manga?A. Baranwal, M. Kataria, N. Agrawal, Y.S. Rawat, S. Vyas
SynSpill: Improved Industrial Spill Detection With Synthetic DataA. Baranwal, A. Mueez, J. Voelker, G. Bhatia, S. Vyas
MolSight: Molecular Property Prediction with ImagesA. Baranwal, A. Gupta, Y.S. Rawat, S. Vyas
The Dark Regulome: Disentangling Predictability from Regulation in Genomic Foundation ModelsC. Baranwal, A. Baranwal, L.N. Tandon