It provides a drag-and-drop interface that allows users to easily create complex image processing workflows without having to write any code.
You can achieve personalized image automation by simply combining different function blocks (such as image editing functions and AI models) as needed.
This tool mainly solves the problem of batch processing pictures encountered in the e-commerce field.
ML Blocks allows users to create custom image processing workflows that can handle multi-step image generation or analysis pipelines, using graph-based workflows. Users only need to connect several blocks in order, such as background removal-cropping-AI upsampling, and they can get a complete image processing workflow in a few minutes.
Main functions:
- Generate images: Use AI models such as Stable Diffusion to generate or draw images.
- Edit image: Provides editing functions such as cropping, resizing, recoloring, etc. to modify the image.
- Analyze images: Use detection or segmentation models to extract data from images.
Practical application examples:
Blur specific areas of the image based on hints: Traditional methods require using the DINO model to generate bounding boxes around the objects mentioned in the hints, then using a segmentation model like Segment Anything to generate masks for these areas, and finally using the Pillow or OpenCV library to write blurring functions to blur the masked areas.
With ML Blocks, users only need to connect segmentation, masking, and fuzzy blocks to complete the workflow in 2 minutes.
You can also automatically generate banner images for blog posts or tweets, remove objects in the images based on prompts, remove backgrounds, and create new backgrounds with AI.
Portal:https://mlblocks.com
Working principle:https://blog.mlblocks.com/p/what-on-earth-is-ml-blocks
Video: