Simple SAM2 TKinter GUI

Overview

The SAM2 TKinter Gui is a Python-based application designed for video processing using the SAM2 (Segment-Anything Model 2) model. The application provides functionality for inpainting videos by removing objects, setting mask colors, and applying diffusion-based inpainting techniques. The app features a user-friendly graphical interface built with Tkinter.

Features

• Video Playback Controls: Play, pause, and stop video playback with a progress slider.
• Frame Annotation: Add positive or negative prompts to frames by clicking on the image.
• Object Removal: Remove objects from videos using color-based, lama, or stable-diffusion-based inpainting.
• Mask Management: Save segmentation masks.
• Video Saving: Save the processed video with inpainted frames.
• Configuration: Customize inpainting settings through the GUI.

Requirements

• Python >= 3.10
• Required Python libraries:
  • yaml
  • cv2 (OpenCV)
  • numpy
  • tkinter
  • torch
  • PIL (Pillow)
  • diffusers
  • tqdm
  • Segment-Anything 2
  • Simple-Lama-Inpainting

Usage

Modify config.yaml and change the checkpoint path, then run

python tkapp.py

You can load a video or a folder containing all the frame images of the video. Use left-click to add a positive point prompt and right-click to add a negative prompt. Run "Process" so that SAM2 can generate the masks for the chosen object. Use either color inpainting, LAMA, or stable-diffusion inpainting to inpaint the video.

Example

Background Removements

color_demo.mp4

Object Removements using LAMA

lama_demo.mp4

Reference

1. Segment-Anything 2: https://github.com/facebookresearch/segment-anything-2
2. LAMA: https://github.com/advimman/lama
3. Simple-LAMA-inpainting: https://github.com/enesmsahin/simple-lama-inpainting