LaMa
Large Mask inpainting model (LaMa) that employs Fourier convolutions and a large receptive field mechanism to leverage global semantic information for high-quality filling and restoration of large occluded regions in images.
Timeline (last 90 days)
Context awareness means that an inpainting model, when generating new content, simultaneously sees the original pixels outside the mask and infers from them the lighting, perspective and texture that the region should have; the encoder takes the entire image together with the mask as input.
Deep learning-based inpainting models such as NVIDIA's Partial Convolution and LaMa (Large Mask inpainting) can leverage global semantic information to generate visually more coherent fill results.
All Facts (2)
Context awareness means that an inpainting model, when generating new content, simultaneously sees the original pixels outside the mask and infers from them the lighting, perspective and texture that the region should have; the encoder takes the entire image together with the mask as input.
50%UnverifiedDeep learning-based inpainting models such as NVIDIA's Partial Convolution and LaMa (Large Mask inpainting) can leverage global semantic information to generate visually more coherent fill results.
50%