ImageNet-21K Pretraining for the MassesDownload PDF

Published: 29 Jul 2021, Last Modified: 22 Oct 2023NeurIPS 2021 Datasets and Benchmarks Track (Round 1)Readers: Everyone
Keywords: ImageNet21K, pretraining, SOTA, semantic softmax, single-label, multi-label, downstream
TL;DR: We present a novel end-to-end scheme for high-quality efficient pretraining on ImageNet-21K dataset
Abstract: ImageNet-1K serves as the primary dataset for pretraining deep learning models for computer vision tasks. ImageNet-21K dataset, which is bigger and more diverse, is used less frequently for pretraining, mainly due to its complexity, low accessibility, and underestimation of its added value. This paper aims to close this gap, and make high-quality efficient pretraining on ImageNet-21K available for everyone. Via a dedicated preprocessing stage, utilization of WordNet hierarchical structure, and a novel training scheme called semantic softmax, we show that various models significantly benefit from ImageNet-21K pretraining on numerous datasets and tasks, including small mobile-oriented models. We also show that we outperform previous ImageNet-21K pretraining schemes for prominent new models like ViT and Mixer. Our proposed pretraining pipeline is efficient, accessible, and leads to SoTA reproducible results, from a publicly available dataset. The training code and pretrained models are available at:
Supplementary Material: zip
Contribution Process Agreement: Yes
Author Statement: Yes
Community Implementations: [![CatalyzeX](/images/catalyzex_icon.svg) 1 code implementation](
8 Replies