Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
[ICLR 2021, Spotlight] Large Scale Image Completion via Co-Modulated Generative Adversarial Networks
| Date | Stars |
|---|---|
| 2026-07-24 | 472 |
| 2026-07-25 | 472 |
| 2026-07-28 | 472 |
| 2026-07-30 | 472 |
| 2026-08-06 | 472 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Large Scale Image Completion via Co-Modulated Generative Adversarial Networks, ICLR 2021 (Spotlight) ### [Demo (Unofficial)](https://www.microsoft.com/en-us/ai/ai-lab-CoModGAN) | [Paper](https://openreview.net/pdf?id=sSjqmfsk95O) <img src="imgs/demo.gif"/> **[NEW!]** Another [unofficial demo](https://www.microsoft.com/en-us/ai/ai-lab-CoModGAN) is available! **[NOTICE]** Our web demo will be closed recently. Enjoy the last days! **[NEW!]** Time to play with our [interactive web demo](http://comodgan.ml)! *Numerous task-specific variants of conditional generative adversarial networks have been developed for image completion. Yet, a serious limitation remains that all existing algorithms tend to fail when handling **large-scale missing regions**. To overcome this challenge, we propose a generic new approach that bridges the gap between image-conditional and recent modulated unconditional generative architectures via **co-modulation** of both conditional and stochastic style representations. Also, due to the lack of good quantitative metrics for image completion, we propose the new **Paired/Unpaired Inception Discriminative Score (P-IDS/U-IDS)**, which robustly measures the perceptual fidelity of inpainted images compared to real images via linear separability in a feature space. Experiments demonstrate superior performance in terms of both quality and diversity over state-of-the-art methods in free-form image completion and easy generalization to image-to-image translation.* <img src="imgs/grid-main.jpg" width="1000px"/> Large Scale Image Completion via Co-Modulated Generative Adversarial Networks<br> [Shengyu Zhao](https://scholar.google.com/citations?user=gLCdw70AAAAJ), [Jonathan Cui](https://www.linkedin.com/in/jonathan-cui-110b211a6/), Yilun Sheng, Yue Dong, Xiao Liang, Eric I Chang, Yan Xu<br> Tsinghua University and Microsoft Research<br> [arXiv](http://arxiv.org/abs/2103.10428) | [OpenReview](https://openreview.net/pdf?id=sSjqmfsk95O) ## Overview This repo is implemented upon and has the same dependencies as the official [StyleGAN2 repo](https://github.com/NVlabs/stylegan2). We also provide a [Dockerfile](https://github.com/zsyzzsoft/co-mod-gan/blob/master/Dockerfile) for Docker users. This repo currently supports: - [x] Large scale image completion experiments on FFHQ and Places2 - [x] Image-to-image translation experiments on Edges2Shoes and Edges2Handbags - [ ] Image-to-image translation experiments on COCO-Stuff - [x] Evaluation code of *Paired/Unpaired Inception Discriminative Score (P-IDS/U-IDS)* ## Datasets - FFHQ dataset (in TFRecords format) can be downloaded following the [StyleGAN2 repo](https://github.com/NVlabs/stylegan2). - Places2 dataset can be downloaded in [this website](http://places2.csail.mit.edu/download.html) (Places365-Challenge 2016 high-resolution images, [training set](http://data.csail.mit.edu/places/places365/train_large_places365challenge.tar) and [validation set](http://data.csail.mit.edu/places/places365/val_large.tar)). The raw images should be converted into TFRecords using `dataset_tools/create_from_images.py` with `--shuffle --compressed`. - Edges2Shoes and Edges2Handbags datasets can be downloaded following the [pix2pix repo](https://github.com/phillipi/pix2pix). The raw images should be converted into TFRecords using `dataset_tools/create_from_images.py` with `--shuffle --pix2pix`. - To prepare a custom dataset, please use `dataset_tools/create_from_images.py`, which will automatically center crop and resize your images to the specified resolution. You only need to specify `--val-image-dir` for testing purpose. ## Training The following script is for training on FFHQ. It will split 10k images for validation. We recommend using 8 NVIDIA Tesla V100 GPUs for training. Training at 512x512 resolution takes about 1 week. ```bash python run_training.py --data-dir=DATA_DIR --dataset=DATASET --metrics=ids10k --mirror-augment --num-gpus=8 ``` The following script is for trainin
Excerpt of 7,285 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:8292d7043b7acb74, topic:tensorflow