Unsupervised representation learning task trained from pairs of images showing the same scene from different viewpoints.
The expohedron is a polytope whose points represent all achievable exposures of items for a Position Based Model (PBM).
Code for running our FIRe model , based solely on mid-level features that we call super-features.
Code repository for paper: What do compressed multilingual machine translation models forget?
A method that improves dense 2D image feature extractors when the latter are applied to the analysis of multiple images reconstructible as a 3D scene.
Model trained by mimicking the BERT algorithm from the natural language processing community.
An auto-regressive transformer-based approach which internally compresses human motion into quantized latent sequences.
A dataset pairing 3D human poses with both automatically generated and human-written descriptions.
Covers more than 10K language pairs, achieves competitive results with M2M-100 while being much smaller and faster.
A Pytorch codebase for research to replicate the CVPR22 paper.
Official repo for the NeurIPS 2022 paper.
These prompted datasets to benchmark the ability of a model to perform completely unseen tasks specified in natural language.
These prompted datasets to benchmark the ability of a model to perform completely unseen tasks specified in natural language.
This repository contains the official implementation of the 3DV 2021 paper.
A general framework for imposing constraints on samples of pretrained language models
Five new indoor datasets with over 130K images.
Publications concern efficient inference, continual learning, unsupervised NMT and domain adaptation.
Dataset that allows exploration of cross-modal retrieval where images contain scene-text instances.
A method that is simple, easy to implement and train and of broad applicability.
Code repository for the ImageNet-CoG Benchmark introduced in the paper ICCV 2021 paper.
Relies strongly on the kapture format for data representation and manipulation.
Code implementing the model introduced in Learning to Rank Images with Cross-Modal Graph Convolutions (ECIR’20).
A novel, efficient model for whole-body 3D pose estimation (including bodies, hands and faces), trained by mimicking the output of hand-, body- and face-pose experts.