Projects

erAIser

Video InpaintingComputer VisionDeep Learning

erAIser removes unwanted objects from videos end to end. The pipeline has three stages, each powered by a deep learning model: video object segmentation (VOSNet), video inpainting (VINet), and re-animation of the removed object’s motion with new imagery (AANet). My role was implementing the VINet-based inpainting stage.

Code on GitHub

Result

Before inpainting.
After inpainting.