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

This repository contains the official implementation code of the paper Transformer-based Feature Reconstruction Network for Robust Multimodal Sentiment Analysis

A Research-oriented Federated Learning Library and Benchmark Platform for Graph Neural Networks. Accepted to ICLR'2021 - DPML and MLSys'21 - GNNSys workshops.

VACA: Designing Variational Graph Autoencoders for Interventional and Counterfactual Queries

PAWS 🐾 Predicting View-Assignments with Support Samples

Detectorch - detectron for PyTorch

[2021][ICCV][FSNet] Full-Duplex Strategy for Video Object Segmentation

Generic template to bootstrap your PyTorch project with PyTorch Lightning, Hydra, W&B, and DVC.

Official implementation of "OpenPifPaf: Composite Fields for Semantic Keypoint Detection and Spatio-Temporal Association" in PyTorch.

Python implementation of Wu et al (2018)'s registration fusion

Python package to generate image embeddings with CLIP without PyTorch/TensorFlow

🥇 LG-AI-Challenge 2022 1위 솔루션 입니다.

PyTorch code for DriveGAN: Towards a Controllable High-Quality Neural Simulation

A Decentralized Omnidirectional Visual-Inertial-UWB State Estimation System for Aerial Swar.

Human motion synthesis using Unity3D

Image data augmentation scheduler for albumentations transforms

Algorithmic trading using machine learning.

Implementation of ECCV20 paper: the devil is in classification: a simple framework for long-tail object detection and instance segmentation

A Demo server serving Bert through ONNX with GPU written in Rust with <3

Poplar implementation of "Bundle Adjustment on a Graph Processor" (CVPR 2020)

PySOT - SenseTime Research platform for single object tracking, implementing algorithms like SiamRPN and SiamMask.