MIMIC: Masked Image Modeling with Image Correspondences

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Marathe, Kalyani, Bigverdi, Mahtab, Khan, Nishat, Kundu, Tuhin, Howe, Patrick, S, Sharan Ranjit, Bhattad, Anand, Kembhavi, Aniruddha, Shapiro, Linda G., Krishna, Ranjay
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911877529862144
author Marathe, Kalyani
Bigverdi, Mahtab
Khan, Nishat
Kundu, Tuhin
Howe, Patrick
S, Sharan Ranjit
Bhattad, Anand
Kembhavi, Aniruddha
Shapiro, Linda G.
Krishna, Ranjay
author_facet Marathe, Kalyani
Bigverdi, Mahtab
Khan, Nishat
Kundu, Tuhin
Howe, Patrick
S, Sharan Ranjit
Bhattad, Anand
Kembhavi, Aniruddha
Shapiro, Linda G.
Krishna, Ranjay
contents Dense pixel-specific representation learning at scale has been bottlenecked due to the unavailability of large-scale multi-view datasets. Current methods for building effective pretraining datasets heavily rely on annotated 3D meshes, point clouds, and camera parameters from simulated environments, preventing them from building datasets from real-world data sources where such metadata is lacking. We propose a pretraining dataset-curation approach that does not require any additional annotations. Our method allows us to generate multi-view datasets from both real-world videos and simulated environments at scale. Specifically, we experiment with two scales: MIMIC-1M with 1.3M and MIMIC-3M with 3.1M multi-view image pairs. We train multiple models with different masked image modeling objectives to showcase the following findings: Representations trained on our automatically generated MIMIC-3M outperform those learned from expensive crowdsourced datasets (ImageNet-1K) and those learned from synthetic environments (MULTIVIEW-HABITAT) on two dense geometric tasks: depth estimation on NYUv2 (1.7%), and surface normals estimation on Taskonomy (2.05%). For dense tasks which also require object understanding, we outperform MULTIVIEW-HABITAT, on semantic segmentation on ADE20K (3.89%), pose estimation on MSCOCO (9.4%), and reduce the gap with models pre-trained on the object-centric expensive ImageNet-1K. We outperform even when the representations are frozen, and when downstream training data is limited to few-shot. Larger dataset (MIMIC-3M) significantly improves performance, which is promising since our curation method can arbitrarily scale to produce even larger datasets. MIMIC code, dataset, and pretrained models are open-sourced at https://github.com/RAIVNLab/MIMIC.
format Preprint
id arxiv_https___arxiv_org_abs_2306_15128
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle MIMIC: Masked Image Modeling with Image Correspondences
Marathe, Kalyani
Bigverdi, Mahtab
Khan, Nishat
Kundu, Tuhin
Howe, Patrick
S, Sharan Ranjit
Bhattad, Anand
Kembhavi, Aniruddha
Shapiro, Linda G.
Krishna, Ranjay
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Dense pixel-specific representation learning at scale has been bottlenecked due to the unavailability of large-scale multi-view datasets. Current methods for building effective pretraining datasets heavily rely on annotated 3D meshes, point clouds, and camera parameters from simulated environments, preventing them from building datasets from real-world data sources where such metadata is lacking. We propose a pretraining dataset-curation approach that does not require any additional annotations. Our method allows us to generate multi-view datasets from both real-world videos and simulated environments at scale. Specifically, we experiment with two scales: MIMIC-1M with 1.3M and MIMIC-3M with 3.1M multi-view image pairs. We train multiple models with different masked image modeling objectives to showcase the following findings: Representations trained on our automatically generated MIMIC-3M outperform those learned from expensive crowdsourced datasets (ImageNet-1K) and those learned from synthetic environments (MULTIVIEW-HABITAT) on two dense geometric tasks: depth estimation on NYUv2 (1.7%), and surface normals estimation on Taskonomy (2.05%). For dense tasks which also require object understanding, we outperform MULTIVIEW-HABITAT, on semantic segmentation on ADE20K (3.89%), pose estimation on MSCOCO (9.4%), and reduce the gap with models pre-trained on the object-centric expensive ImageNet-1K. We outperform even when the representations are frozen, and when downstream training data is limited to few-shot. Larger dataset (MIMIC-3M) significantly improves performance, which is promising since our curation method can arbitrarily scale to produce even larger datasets. MIMIC code, dataset, and pretrained models are open-sourced at https://github.com/RAIVNLab/MIMIC.
title MIMIC: Masked Image Modeling with Image Correspondences
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2306.15128