A video scene detection algorithm is designed to detect a variety of different scenes within a video

Last update: Jan 04, 2022

Overview

Scene-Change-Detection

The detection of scenes change is a simple problem that human beings face, but it gets much harder to handle autonomously a device that generally includes complex calculations and algorithms.

A video scene detection algorithm is designed to detect a variety of different scenes within a video. There is a very simple definition for a scene: It is a series of logically and chronologically related shots taken in a specific order to depict an over-arching concept or story. The identification of video scenes, in many video analysis applications, is a crucial pre-processing step. A dataset for video scene detection known as the Open Video Scene Detection (OVSD) dataset has been provided in order to evaluate algorithms for video scene detection. Videos in the dataset have an open-source nature, which makes them an ideal product to be used by academics, as well as industry researchers alike.

DATASET

Dataset 2012 - In the dataset, there are six video categories, and in each category, there are four to six video sequences
IBM video Scene Change Detection
A dataset for video scene detection known as the Open Video Scene Detection (OVSD) dataset has been provided in order to evaluate algorithms for video scene detection.

MODEL

VGG16 was used for this project.VGG16 is a convolutional neural network model proposed by K. Simonyan and A. Zisserman from the University of Oxford in the paper “Very Deep Convolutional Networks for Large- Scale Image Recognition”. The model achieves 92.7% top-5 test accuracy in ImageNet, which is a dataset of over 14 million images belonging to 1000 classes.

LIBRARIES USED

Numpy for array manipulation
OpenCV (cv2) for Image Augmentation
Keras for building the Neural Network
Matplotlib for plotting visuals

COMPILATION USED

Loss function selected is sparse categorical cross-entropy
Optimizer selected is Adam
Validation metric chosen is accuracy

Training

No of epochs = 5
Batch size = 1

REPORT

https://drive.google.com/file/d/1cwoP5cRJ5D76PvHV_WjCDRoSJIf0h9du/view?usp=sharing

COLLABORATORS

Neel kumar arya and Ashish Vidyarthi

A video scene detection algorithm is designed to detect a variety of different scenes within a video

Related tags

Overview

Scene-Change-Detection

DATASET

MODEL

LIBRARIES USED

COMPILATION USED

Training

REPORT

COLLABORATORS

License

Owner

Unconstrained Text Detection with Box Supervisionand Dynamic Self-Training

Source code of AAAI 2022 paper "Towards End-to-End Image Compression and Analysis with Transformers".

Machine learning Bot detection technique, based on United States election dataset

Official implementation of ACTION-Net: Multipath Excitation for Action Recognition (CVPR'21).

[内测中]前向式Python环境快捷封装工具，快速将Python打包为EXE并添加CUDA、NoAVX等支持。

Implementation of "Semi-supervised Domain Adaptive Structure Learning"

PySLM Python Library for Selective Laser Melting and Additive Manufacturing

A Dynamic Residual Self-Attention Network for Lightweight Single Image Super-Resolution

Framework for Spectral Clustering on the Sparse Coefficients of Learned Dictionaries

MEAL V2: Boosting Vanilla ResNet-50 to 80%+ Top-1 Accuracy on ImageNet without Tricks

Liecasadi - liecasadi implements Lie groups operation written in CasADi

Seeing if I can put together an interactive version of 3b1b's Manim in Streamlit

Notes taking website build with Docker + Django + React.

Show-attend-and-tell - TensorFlow Implementation of "Show, Attend and Tell"

PyTorch implementation DRO: Deep Recurrent Optimizer for Structure-from-Motion

Dahua Camera and Doorbell Home Assistant Integration

SeMask: Semantically Masked Transformers for Semantic Segmentation.

Self-Supervised Image Denoising via Iterative Data Refinement

Tensorflow implementation of "BEGAN: Boundary Equilibrium Generative Adversarial Networks"

The Noise Contrastive Estimation for softmax output written in Pytorch