https://github.com/shreshth-112/video-summarization-using-keyframe-extraction

Built a model to create highlights/summary of given video. The results of this study shows that, with a remarkable similarity index(SSIM) of 98%, the recommended technique is quite successful in choosing keyframes that are both educational and distinctive from the original movie
https://github.com/shreshth-112/video-summarization-using-keyframe-extraction

k-means-clustering vae-implementation

Last synced: 12 days ago
JSON representation

Built a model to create highlights/summary of given video. The results of this study shows that, with a remarkable similarity index(SSIM) of 98%, the recommended technique is quite successful in choosing keyframes that are both educational and distinctive from the original movie

• Host: GitHub
• URL: https://github.com/shreshth-112/video-summarization-using-keyframe-extraction
• Owner: Shreshth-112
• Created: 2024-05-18T05:41:36.000Z (6 months ago)
• Default Branch: main
• Last Pushed: 2024-05-18T06:00:01.000Z (6 months ago)
• Last Synced: 2024-05-19T06:32:48.605Z (6 months ago)
• Topics: k-means-clustering, vae-implementation
• Language: Jupyter Notebook
• Homepage:
• Size: 112 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# ABSTRACT

In order to choose the keyframes from a video sequence that best convey its information, our project on keyframe extraction offers a unique method that blends autoencoder-based feature extraction with nested k-means clustering. Keyframe selection and extraction from videos is a crucial operation since it may assist to drastically minimise the amount of data that has to be processed or preserved and enables viewers to get a broad idea of the content without having to watch the full sequence. Our suggested technique makes use of a variational autoencoder (VAE), a form of autoencoder that has a probabilistic model that makes feature extraction more efficient and dependable. Because VAEs can more correctly capture the distribution of the data, which results in more accurate feature extraction, they are regarded to be an improvement over typical autoencoders. In order to find the most representative frames or keyframes, the VAE-based feature extraction approach is utilised to extract significant characteristics from the video frames. The suggested technique makes use of nested k-means clustering to determine which frames or keyframes are the most representative. The layered clustering technique enables the selection of keyframes that are more indicative of the particular video features in which the viewer is interested. It also allows for finer-grained grouping discoveries. This method makes keyframe extraction more accurate and efficient, which is especially helpful for applications like video summary, retrieval, and indexing. The effectiveness of the suggested technique on different datasets is demonstrated by experimental findings.

## example -

![Alt text](https://github.com/Shreshth-112/Video-summarization-using-keyframe-extraction/assets/136225408/b1ec40bf-a6f4-4f73-aea7-37d41a59228d)