L-verse: Bidirectional Generation Between Image and Text

Taehoon Kim, Gwangmo Song, Sihaeng Lee, Sangyun Kim, Yewon Seo, Soonyoung Lee, Seung Hwan Kim, Honglak Lee, Kyunghoon Bae | Paper | Pretrained Models

LG AI Research

Abstract

Far beyond learning long-range interactions of natural language, transformers are becoming the de-facto standard for many vision tasks with their power and scalabilty. Especially with cross-modal tasks between image and text, vector quantized variational autoencoders (VQ-VAEs) are widely used to make a raw RGB image into a sequence of feature vectors. To better leverage the correlation between image and text, we propose L-Verse, a novel architecture consisting of feature-augmented variational autoencoder (AugVAE) and bidirectional auto-regressive transformer (BiART) for text-to-image and image-to-text generation. Our AugVAE shows the state-of-the-art reconstruction performance on ImageNet1K validation set, along with the robustness to unseen images in the wild. Unlike other models, BiART can distinguish between image (or text) as a conditional reference and a generation target. L-Verse can be directly used for image-to-text or text-to-image generation tasks without any finetuning or extra object detection frameworks. In quantitative and qualitative experiments, L-Verse shows impressive results against previous methods in both image-to-text and text-to-image generation on MS-COCO Captions. We furthermore assess the scalability of L-Verse architecture on Conceptual Captions and present the initial results of bidirectional vision-language representation learning on general domain.

Requirements

pip install -r requirements.txt

Data Preparation

Place any image dataset with ImageNet-style directory structure (at least 1 subfolder) to fit the dataset into pytorch ImageFolder.

Training AugVAEs

You can easily test train_vae.py with randomly generated fake data.

python train_vae.py --fake_data

For actual training, provide a config file:

python train_vae.py --configs [config_file]

Please refer to example config files in configs. You first need to train AugVAE-ML before training AugVAE-SL.

Pretrained models

We provide the AugVAE-SL pretrained weight on ImageNet dataset. Google Drive

Training BiART

Our BiART implementation is based on Andrej Karpathy’s awesome GPT implementation in minGPT. We provide custom BiART implementation with segment embedding in cond_transformer.py and mingpt.py.

License

This project is distributed under MIT license.

Copyright (c) 2021-present LG AI Research.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in
all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.