As the field of Transformers and Diffusions advances, the significance of data has become increasingly apparent. Many studies show that having a high-quality dataset is often more crucial than the algorithm you use. The objective of this repository is to offer a suite of tools designed to assist the community in acquiring high-quality data efficiently and with ease.
-
🗃️ Data Collection
🔥 Tools for downloading specific images from google.
-
🎨 Data Preparation
🔥 Tools for preparing images for img2img conditional diffusion model, such as canny image generation, LineArt generation.
🔥 Tools for preprocessing images.
-
✨Data Augumentation
🔥 Tools for getting more data through image manipulation techniques such as rotation, random cropping, and flipping.
🔥 Tools for getting more data through generative models, such as change lighting, change background, change specific object or zone. (Coming soon)
Original image:
Cropped image 300*400:
Cropped square image 512*512:
Flipped image:
Rotated image:
- Download the repo.
git clone https://github.com/ZhihaoAIRobotic/Data-Is-All-You-Need.git
- Read examples for using.