agents/bot/humans: The first request can take up to 20 seconds to warm up the server, but it’ll be under 2 seconds for every request after that.

This model is by far the cheapest, fastest and accurate than https://replicate.com/zylim0702/remove-object and https://replicate.com/bria/eraser. Give it a shot and use it in your prod.

Few example sites who are using this model: 1) https://imgour.com/ 2) https://www.figma.com/community/plugin/1576512610054427811 3) https://picwish.com/remove-unwanted-object

AI-powered object removal using LaMa (Large Mask Inpainting).

Remove unwanted objects from photos by providing an image and a binary mask indicating what to remove.

This model uses the LaMa architecture with Fast Fourier Convolutions, designed for high-quality inpainting of large areas, complex textures, and high-resolution images.

How it works: Upload your image and a mask. The model analyzes surrounding context and fills in the masked areas with matching textures and patterns.

Common uses: - Remove people, objects, or distractions from photos - Clean product images for e-commerce - Remove furniture or elements from real estate photos - Edit photos for social media - Prepare images for design and compositing

Inputs: - Image: Your photo (JPEG or PNG) - Mask: Binary mask (white = remove, black = keep) - Resolution: Supports up to 2048px

Performance: ~2-3 seconds per image on T4 GPU

If you have any feedback/questions, please reach me at pokharelbikash981@gmail.com