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

Conditional Generative Adversarial Networks (CGAN) for Mobility Data Fusion

Semi-Supervised Learning, Object Detection, ICCV2021

Spatial Sparse Convolution Library

A python package simulating the quasi-2D pseudospin-1/2 Gross-Pitaevskii equation with NVIDIA GPU acceleration.

Labelbox is the fastest way to annotate data to build and ship artificial intelligence applications

Gym for multi-agent reinforcement learning

Tree Nested PyTorch Tensor Lib

Collection of generative models in Pytorch version.

PyStan, a Python interface to Stan, a platform for statistical modeling. Documentation: https://pystan.readthedocs.io

[EMNLP 2021] MuVER: Improving First-Stage Entity Retrieval with Multi-View Entity Representations

An intuitive library to extract features from time series

Sequence Modeling with Structured State Spaces

Deep Unsupervised 3D SfM Face Reconstruction Based on Massive Landmark Bundle Adjustment.

[NeurIPS 2021] Code for Unsupervised Learning of Compositional Energy Concepts

Julia package for contraction of tensor networks, based on the sweep line algorithm outlined in the paper General tensor network decoding of 2D Pauli codes

Experiments for Operating Systems Lab (ETCS-352)

Black box hyperparameter optimization made easy.

Computationally Efficient Optimization of Plackett-Luce Ranking Models for Relevance and Fairness

G-NIA model from "Single Node Injection Attack against Graph Neural Networks" (CIKM 2021)

[ICCV2021] Learning to Track Objects from Unlabeled Videos