Completed
Vision, Language and Visual Retrieval
Multimodal methods connecting images and language: large-scale visual retrieval, semantic art understanding, and image caption generation.
Computer Vision
Image and video understanding, geometry, and recognition.
Projects
Completed
Self-Supervised Monocular Depth Estimation
Learning to predict depth from single images without ground-truth supervision, with a focus on dynamic scenes and challenging conditions.
Completed
3D Reconstruction and Multi-View Stereo
Recovering accurate 3D shape from collections of images, using multi-view stereo, volumetric graph-cuts, and probabilistic depth-map fusion.
Publications
- BaseBoostDepth: Exploiting Larger Baselines For Self-supervised Monocular Depth EstimationarXiv (Cornell University) · 2024
- Self-supervised Monocular Depth Estimation: Let’s Talk About The WeatherarXiv (Cornell University) · 2023
- Dyna-DM: Dynamic Object-aware Self-supervised Monocular Depth Maps2023
- Dyna-DM: Dynamic Object-aware Self-supervised Monocular Depth MapsarXiv (Cornell University) · 2022
- Variational Recurrent Sequence-to-Sequence Retrieval for Stepwise IllustrationLecture notes in computer science · 2020
- Learning non-metric visual similarity for image retrievalImage and Vision Computing · 2019
- How to Read Paintings: Semantic Art Understanding with Multi-modal RetrievalLecture notes in computer science · 2019
- A Deep Learning Approach to Automatic Caption Generation for News ImagesAston Publications Explorer (Aston University) · 2019
- Asymmetric Spatio-Temporal Embeddings for Large-Scale Image-to-Video RetrievalAston Publications Explorer (Aston University) · 2018
- Neural Caption Generation for News Images2018
- How to Read Paintings: Semantic Art Understanding with Multi-Modal RetrievalarXiv (Cornell University) · 2018
- Dress Like a Star: Retrieving Fashion Products from Videos2017
- Large-Scale Data for Multiple-View StereopsisInternational Journal of Computer Vision · 2016
- Our 3D Vision Data-Sets in the MakingTechnical University of Denmark, DTU Orbit (Technical University of Denmark, DTU) · 2015
- Large Scale Multi-view Stereopsis Evaluation2014
- A Generative Model for Online Depth FusionLecture notes in computer science · 2012
- WITHDRAWN: Video-based, real-time multi-view stereoImage and Vision Computing · 2012
- Video-based, real-time multi-view stereoImage and Vision Computing · 2011
- Automatic Object Segmentation from Calibrated Images2011
- Live 3D shape reconstruction, recognition and registration2011
- Shape from Photographs: A Multi-view Stereo PipelineStudies in computational intelligence · 2010
- Practical 3D Reconstruction Based on Photometric StereoStudies in computational intelligence · 2010
- Using Multiple Hypotheses to Improve Depth-Maps for Multi-View StereoLecture notes in computer science · 2008
- Automatic 3D object segmentation in multiple views using volumetric graph-cutsImage and Vision Computing · 2008
- Multiview Stereo via Volumetric Graph-Cuts and Occlusion Robust Photo-ConsistencyIEEE Transactions on Pattern Analysis and Machine Intelligence · 2007
- Probabilistic visibility for multi-view stereo2007
- Automatic 3D Object Segmentation in Multiple Views using Volumetric Graph-Cuts2007
- Reconstructing relief surfacesImage and Vision Computing · 2007
- Lighting-Up Geometry: Accurate 3D Modelling of Museum Artifacts with a Torch and a CameraCambridge University Engineering Department Publications Database · 2006
- Multi-View Stereo via Volumetric Graph-Cuts2005
- Using frontier points to recover shape, reflectance and illumination2005
- Reconstructing Relief Surfaces2004
- Bayesian Stochastic Mesh Optimization for 3D reconstruction2003