FrameSift - Advanced Keyframe Detection with News Analysis & Text Extraction

A comprehensive desktop application for intelligent video keyframe extraction featuring specialized News Content Detection and IVP-based Text Extraction. Built with advanced Image and Video Processing (IVP) algorithms and modern Python GUI frameworks.

Overview

FrameSift is a powerful keyframe extraction tool that combines traditional video processing with specialized news content analysis and intelligent text extraction. The application achieves 70-80% frame reduction while providing advanced features for analyzing news broadcasts, extracting text from keyframes, and applying comprehensive IVP techniques.

New Features (2025 Update)

News Content Detection: Specialized analysis for news broadcasts and clips
Text Extraction: IVP-based OCR with selective processing
Intelligent Export: Text extraction only on selected keyframes
Advanced IVP: Comprehensive image processing pipeline
Smart Thermal Management: Automatic fan control during processing

Key Features

News Content Detection

Intelligent Transition Detection: Automatically finds content changes in news broadcasts
Multi-Modal Analysis: Combines face detection, text analysis, and scene change detection
ROI-Based Processing: Focuses on regions where news content typically changes
Transition Classification: Categorizes changes as headline, person, or scene transitions
Confidence Scoring: Provides reliability metrics for each detected transition
Export Integration: Seamless export with detailed transition reports

Advanced Text Extraction (IVP-Based)

Selective OCR Processing: Text extraction only on exported keyframes (not during detection)
Multi-Engine OCR: EasyOCR (deep learning) + Tesseract (traditional) for maximum accuracy
IVP Enhancement Pipeline: Comprehensive image preprocessing for optimal text detection
Text Region Classification: Automatically categorizes text as headlines, tickers, captions, or content
Visual Annotations: Bounding boxes and confidence scores for detected text regions
JSON Export: Structured text data with coordinates, confidence, and classification

IVP (Image & Video Processing) Techniques

Image Enhancement: Contrast stretching, histogram equalization, Gaussian filtering
Morphological Operations: Dilation, erosion, opening, closing for text detection
Edge Detection: Canny edge detector for boundary enhancement
Image Segmentation: Otsu thresholding, region-based segmentation
Quality Analysis: Blur detection using Laplacian variance
Multi-Algorithm Fusion: Combines difference, histogram, and quality metrics

Desktop Application Features

Modern GUI: Intuitive tabbed interface with real-time progress tracking
Thermal Management: Automatic fan control and temperature monitoring
Visual Analytics: Interactive plots and comprehensive statistics
Flexible Export: Multiple output formats with customizable options
Parameter Presets: Conservative, Balanced, and Aggressive processing modes
Live Preview: Thumbnail gallery with detailed keyframe information

IVP Algorithms & Techniques

Core Video Processing Algorithms

Algorithm	Purpose	Method	Application
Frame Difference	Redundancy removal	Mean Absolute Difference (MAD)	General keyframe extraction
Histogram Comparison	Scene boundary detection	Chi-Square distance	Shot boundary detection
Blur Detection	Quality filtering	Variance of Laplacian	Quality assessment
Face Detection	Person tracking	Haar Cascade Classifiers	News content analysis
ROI Analysis	Content-specific detection	Regional difference metrics	News transition detection

IVP Text Enhancement Pipeline

Stage	Technique	Purpose	Implementation
1. Preprocessing	Grayscale conversion	Simplify processing	Standard OpenCV conversion
2. Enhancement	Contrast stretching	Improve text visibility	Piecewise linear transformation
3. Histogram Processing	Histogram equalization	Optimize contrast distribution	Adaptive enhancement
4. Noise Reduction	Gaussian filtering	Remove image noise	3x3 kernel smoothing
5. Binarization	Otsu thresholding	Create binary text image	Automatic threshold selection
6. Morphological Ops	Opening & Closing	Clean text regions	Structured element operations
7. Edge Detection	Canny edge detector	Enhance text boundaries	Dual threshold edge detection

Multi-Modal Processing Pipeline

Standard Keyframe Extraction

Video Loading → Frame extraction and preprocessing
Algorithm Application → Difference, histogram, and blur analysis
Score Fusion → Combine multiple algorithm outputs
Threshold Application → Apply user-defined selection criteria
Quality Filtering → Remove low-quality frames
Result Generation → Create thumbnails and statistics

News Content Detection

ROI Definition → Define regions for headline, person, and scene analysis
Multi-Modal Analysis → Apply face detection, text change detection, and scene analysis
Temporal Filtering → Ensure minimum gap between transitions
Confidence Scoring → Calculate reliability metrics
Classification → Categorize transition types
Keyframe Extraction → Save transition frames for further analysis

Text Extraction (Selective)

User Selection → Export specific keyframes for text analysis
IVP Enhancement → Apply comprehensive image preprocessing
Multi-Engine OCR → Use EasyOCR and Tesseract for maximum coverage
Text Classification → Categorize regions (headlines, tickers, captions)
Confidence Assessment → Evaluate OCR reliability
Structured Export → Save results in JSON format with visual annotations

Installation

Quick Setup

# Clone repository
git clone <repository-url>
cd "Keyframe Detection"

# Create virtual environment (recommended)
python -m venv venv

# Windows activation:
venv\Scripts\activate

# Linux/Mac activation:
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

Automated Installation

# Run automated installation script
python install.py

System Requirements

Python: 3.7+ (Tkinter included)
Operating System: Windows, macOS, Linux
Memory: 8GB minimum, 16GB recommended for news analysis with text extraction
Storage: 2GB free space for results, models, and temporary files
Display: 1920x1080 recommended for optimal GUI experience
GPU: Optional CUDA support for faster EasyOCR processing
Internet: Required for initial EasyOCR model download (~200MB)

Usage

Launch Application

Option 1: Using the launcher (Recommended)

python launch_gui.py

Option 2: Direct launch

python framesift_gui.py

Option 3: Windows batch file

launch_gui.bat

Application Workflow

Standard Keyframe Extraction

Upload Tab: Select video file and configure basic settings
Parameters Tab: Fine-tune algorithm thresholds or use presets
Process: Run standard keyframe extraction algorithms
Results Tab: View extracted keyframes, statistics, and performance plots
Export: Save keyframes and copy statistics to clipboard

News Content Analysis (Specialized)

News Detection Tab: Upload news broadcast or clip
Configure Thresholds: Set sensitivity for headline, person, and scene changes
Detect Transitions: Run news-specific content analysis
Review Results: View detected transitions with confidence scores
Export Frames: Text extraction happens here - save selected keyframes with OCR analysis
Get Results: Receive JSON files with text extraction data and visual annotations

Text Extraction Features

Selective Processing: Text extraction only runs on keyframes you choose to export
IVP Enhancement: Automatic image preprocessing for optimal OCR results
Multi-Engine OCR: Combines EasyOCR (neural) and Tesseract (traditional) for best accuracy
Intelligent Classification: Automatically categorizes text as headlines, tickers, captions, or content
Structured Output: JSON files with text content, confidence scores, and bounding box coordinates

News Content Detection & Text Extraction

Overview

FrameSift features advanced News Content Detection specifically designed for analyzing news broadcasts, clips, and similar structured content. The system combines computer vision techniques with intelligent text extraction to provide comprehensive analysis of news media.

What It Detects

Transition Type	Description	Detection Method	Use Cases
Headline Changes	Breaking news banners, lower thirds, text overlays	ROI-based difference analysis + text detection	Content analysis, topic tracking
Person Changes	Anchors, reporters, guests appearing/disappearing	Haar cascade face detection	Speaker identification, interview analysis
Scene Changes	Camera cuts, location changes, graphic transitions	Histogram comparison + motion analysis	Shot boundary detection, program structure

Technical Implementation

ROI-Based Analysis

Headline Region: Upper 30% of frame for breaking news and banners
Person Region: Central area optimized for face detection
Scene Analysis: Full-frame histogram and difference metrics
Adaptive Thresholding: User-configurable sensitivity for each region

Multi-Modal Fusion

News Frame → [ROI Extraction] → [Face Detection] → [Text Change Analysis] → [Scene Analysis] → [Confidence Scoring] → [Transition Classification]

Text Extraction Pipeline

When Text Extraction Happens

Fast Detection Phase: News content analysis WITHOUT text extraction (3-5 seconds) Selective Export Phase: Text extraction ONLY on keyframes you choose to export

IVP-Based Text Enhancement

Contrast Stretching: Piecewise linear transformation for better text visibility
Histogram Equalization: Optimize brightness distribution
Gaussian Filtering: Remove noise while preserving text edges
Otsu Thresholding: Automatic binarization for optimal text separation
Morphological Operations: Clean text regions using opening/closing
Canny Edge Detection: Enhance text boundaries for better OCR

OCR Engine Fusion

EasyOCR: Deep learning-based recognition (primary engine)
Tesseract: Traditional OCR (fallback and validation)
Confidence Weighting: Best results from both engines
Duplicate Elimination: Smart merging of overlapping detections

Text Region Classification

Region Type	Location	Characteristics	Example Content
Headlines	Upper 30%, wide regions	Large text, high contrast	"BREAKING NEWS", "WEATHER ALERT"
Tickers	Bottom 20%, continuous	Scrolling text, small font	"STOCK: DOW +2.5% NASDAQ +1.8%"
Captions	Lower half, small regions	Names, locations, credits	"John Smith, Political Analyst"
Content	Anywhere, variable size	General text content	Times, temperatures, other info

Configuration Options

News Detection Thresholds

Headline Threshold (0.1-1.0): Sensitivity to text/banner changes
Person Threshold   (0.1-1.0): Sensitivity to face appearance/disappearance  
Scene Threshold    (0.1-1.0): Sensitivity to general scene transitions
Min Gap           (0.5-10s):  Minimum time between detected transitions

Text Extraction Settings

Confidence Threshold: Minimum OCR confidence to accept text (default: 0.3)
Preprocessing Level: Amount of IVP enhancement applied
OCR Language: Primary language for text recognition (default: English)
Region Filtering: Minimum text region size and aspect ratio

Workflow Example

Step 1: Fast News Detection

Upload news video → Configure thresholds → Run detection → View transitions (3-5 seconds)

Step 2: Selective Text Extraction

Select keyframes → Click "Export Frames" → OCR processing → Get JSON results

Output Formats

Keyframe Export

Image Files: High-quality JPEG keyframes at transition points
Naming Convention: news_keyframe_XXXXXX_[type]_[confidence].jpg
Visual Annotations: Optional bounding boxes around detected text

Text Extraction JSON

{
  "keyframe_path": "exported_keyframe.jpg",
  "frame_number": 125,
  "timestamp": 6.25,
  "transition_type": "headline_change",
  "ocr_method": "EasyOCR + Tesseract",
  "total_confidence": 0.87,
  "text_regions": [
    {
      "text": "BREAKING NEWS",
      "confidence": 0.95,
      "bbox": [47, 16, 310, 43],
      "region_type": "headline"
    }
  ]
}

Export Summary Report

Processing Statistics: Detection parameters, processing time, transition counts
Text Extraction Summary: OCR success rate, confidence distribution
File Manifest: Complete list of exported files with metadata
Quality Metrics: Detection confidence, OCR accuracy statistics

Use Cases & Applications

Professional Applications

Broadcast Analysis: Analyze news program structure and content flow
Content Research: Extract specific segments from long broadcasts
Media Monitoring: Track news coverage and headline changes
Archive Processing: Process historical news footage for searchable content
Academic Research: Study news presentation, bias, and content patterns

Technical Applications

AI Training Data: Generate labeled datasets for ML model training
Content Indexing: Create searchable databases of news content
Analytics: Quantify news content characteristics and patterns
Real-time Processing: Live analysis of news streams and broadcasts

Parameter Configuration

Algorithm Thresholds

Frame Difference Threshold (0.1-1.0)

Lower (0.1-0.3): More sensitive, keeps more frames with subtle changes
Higher (0.6-1.0): Less sensitive, removes more similar frames

Histogram Threshold (0.1-1.0)

Lower (0.1-0.4): Detects only major scene changes
Higher (0.6-0.9): Detects subtle lighting/color changes

Blur Threshold (10-200)

Lower (10-40): Accepts slightly blurry frames
Higher (70-200): Only keeps very sharp frames

Available Presets

Conservative: Higher thresholds, keeps more frames
Balanced: Default settings for most use cases
Aggressive: Lower thresholds, maximum reduction

Example Results

Try the included example videos to see FrameSift Lite in action:

# Generate example videos (optional)
python create_example_videos.py

Video Type	Demonstrates	Expected Reduction
`scene_transitions.mp4`	Shot boundary detection	~75%
`motion_demo.mp4`	Motion analysis	~70%
`quality_variation.mp4`	Blur filtering	~80%
`mixed_content.mp4`	All techniques combined	~75%

Export Options

Save Results: Export processing statistics as JSON
Export Keyframes: Save keyframe images to folder
Copy Statistics: Copy detailed stats to clipboard

Project Structure

FrameSift/
├── Application Launchers
│   ├── framesift_gui.py              # Main Tkinter GUI application
│   ├── launch_gui.py                 # Application launcher with dependency checks
│   ├── launch_gui.bat               # Windows batch launcher
│   └── install.py                   # Automated installation script
│
├── Demo & Testing Scripts
│   ├── demo_video_processor.py       # Command-line processing demo
│   ├── demo_news_detection.py       # News detection demonstration
│   ├── demo_news_text_extraction.py # Text extraction demo
│   ├── test_ocr_synthetic.py        # OCR testing with synthetic images
│   ├── test_selective_extraction.py # Selective text extraction validation
│   └── create_example_videos.py     # Example video generator
│
├── Core Processing Modules
│   ├── modules/
│   │   ├── __init__.py
│   │   ├── ivp_algorithms.py         # Traditional IVP algorithm implementations
│   │   ├── video_processor.py        # Video processing pipeline
│   │   ├── results_generator.py      # Results formatting and export
│   │   ├── news_detector.py          # News content detection & transition analysis
│   │   ├── text_extractor.py         # IVP-based OCR text extraction
│   │   └── simple_fan_control.py     # Thermal management system
│
├── Test Suite
│   ├── tests/
│   │   ├── test_ivp_algorithms.py    # Algorithm unit tests
│   │   ├── test_video_processor.py   # Processing pipeline tests
│   │   ├── test_results_generator.py # Results generation tests
│   │   └── test_news_detection.py    # News detection validation
│
├── Example Content
│   ├── examples/                     # Demo video files
│   │   ├── mixed_content.mp4         # Multi-algorithm demonstration
│   │   ├── scene_transitions.mp4     # Shot boundary detection
│   │   ├── motion_demo.mp4          # Motion analysis showcase
│   │   └── quality_variation.mp4     # Blur filtering example
│   │
│   └── Examples using YouTube Videos/ # Real-world examples with presets
│
├── Output Directories
│   ├── gui_results/                  # Standard keyframe extraction results
│   ├── demo_news_results/           # News detection demonstration output
│   ├── demo_news_text_results/      # Text extraction demo results
│   └── test_selective_extraction/   # Selective processing test output
│
├── Documentation
│   ├── README.md                    # This comprehensive guide
│   ├── ALGORITHMS.md                # Detailed algorithm documentation
│   ├── NEWS_DETECTION_SUMMARY.md   # News detection implementation details
│   ├── TEXT_EXTRACTION_SUMMARY.md  # Text extraction & IVP techniques
│   ├── SELECTIVE_TEXT_EXTRACTION_SUMMARY.md # Selective processing implementation
│   └── SIMPLE_FAN_CONTROL_SUMMARY.md # Thermal management documentation
│
├── Configuration
│   ├── requirements.txt             # Python dependencies (updated with OCR)
│   ├── .gitignore                  # Git ignore rules
│   └── framesift_gui.log           # Application runtime logs
└──

Testing

Unit Tests

# Run all tests
python -m pytest tests/ -v

# Run specific test modules
python -m pytest tests/test_ivp_algorithms.py -v
python -m pytest tests/test_video_processor.py -v
python -m pytest tests/test_results_generator.py -v

# Run with coverage report
python -m pytest tests/ --cov=modules --cov-report=html

Command Line Demo

# Test video processing with demo script
python demo_video_processor.py examples/motion_demo.mp4

Dependencies

Core Computer Vision & Processing

opencv-python>=4.8.1.78    # Video processing, face detection, image operations
numpy>=1.24.3               # Numerical computations and array operations
matplotlib>=3.7.2           # Plotting and data visualization
Pillow>=10.0.1              # Image processing and thumbnail generation

Text Extraction & OCR

pytesseract>=0.3.10         # Traditional OCR engine interface
easyocr>=1.7.0              # Deep learning-based OCR engine

System Management

psutil>=5.9.0               # System monitoring and thermal management

GUI & Standard Libraries

tkinter                     # GUI framework (included with Python standard library)
threading                  # Multi-threaded processing support  
json                        # JSON data handling (standard library)
logging                     # Application logging (standard library)
subprocess                 # System command execution (standard library)

Development & Testing Tools

pytest                      # Unit testing framework (development only)

Installation Commands

# Install all dependencies
pip install -r requirements.txt

# Or install individually
pip install opencv-python numpy matplotlib Pillow pytesseract easyocr psutil

# For development
pip install pytest

External Dependencies (Auto-Downloaded)

EasyOCR Models: ~200MB detection and recognition models (downloaded on first use)
Tesseract: Traditional OCR engine (optional, for backup OCR functionality)
- Windows: Download from GitHub releases
- Linux: sudo apt-get install tesseract-ocr
- macOS: brew install tesseract

Performance & System Requirements

Processing Speed

Standard Keyframe Extraction

Small videos (< 100MB): Real-time processing (1-2x speed)
Medium videos (100MB - 1GB): 2-5x faster than real-time
Large videos (> 1GB): 1-3x real-time depending on resolution

News Detection (Without Text Extraction)

Fast Analysis: 3-6x faster than traditional methods
Real-time Capable: Process 1080p videos at 2-4x speed
Optimized Pipeline: Focus on content-relevant regions only

Text Extraction (Selective Processing)

Per-frame OCR: 2-5 seconds per keyframe (depends on text complexity)
EasyOCR: 1-3 seconds per frame (GPU accelerated when available)
Tesseract: 0.5-1 second per frame (CPU-based)
Selective Advantage: Only process frames you actually need

Memory Usage

Operation	Base RAM	Peak RAM	GPU VRAM
Application Startup	~100MB	~150MB	N/A
Standard Processing	~200MB	~500MB	N/A
News Detection	~300MB	~800MB	N/A
Text Extraction	~500MB	~1.5GB	~2GB (if GPU)
Large Video Processing	~800MB	~2GB	~2-4GB (if GPU)

Supported Formats

Input Video Formats

Recommended: MP4 (H.264), AVI, MOV, MKV Supported: WMV, FLV, WEBM, M4V, 3GP Optimal: 1080p or lower resolution for best performance

Output Formats

Keyframe Images: JPEG (high quality), PNG (optional)
Statistics: JSON, CSV, TXT summary reports
Visualizations: PNG plots, interactive matplotlib figures
Text Data: JSON with full OCR results and annotations
Export Packages: ZIP archives with complete analysis results

Smart Thermal Management

Automatic Fan Control

Processing Detection: Automatically enables high-performance cooling during intensive operations
Power Plan Management: Switches between "Balanced" and "High Performance" modes
Status Monitoring: Real-time fan speed and power plan status in GUI
Auto-Recovery: Returns to normal mode when processing completes

Status Indicators

Status	Description
Fan: Normal	Standard cooling mode, balanced power plan
Fan: HIGH SPEED	Active cooling during processing
Processing	High-performance mode enabled
Cool	System temperature optimal

Cross-Platform Support

Windows: Power plan switching, WMI temperature monitoring
Linux: CPU governor control, thermal zone monitoring
macOS: Power management and thermal control

Performance Optimization Tips

For Best Speed

Use SSD storage for input/output operations
Close other applications during processing
Enable GPU acceleration for EasyOCR (if available)
Process smaller video segments for faster results
Use "Aggressive" presets for maximum frame reduction

For Best Quality

Use "Conservative" presets to keep more frames
Lower blur thresholds to accept slightly blurry frames
Increase news detection thresholds for higher confidence
Process at original resolution (don't downscale)
Use both OCR engines for maximum text extraction accuracy

Memory Management

Large videos: Process in segments if memory limited
Multiple videos: Process sequentially, not simultaneously
GPU memory: Monitor VRAM usage during text extraction
Temp files: Application automatically cleans temporary data

Troubleshooting

Common Issues & Solutions

Issue	Symptoms	Solution
"Cannot open video file"	Error on video load	Check format support (MP4, AVI, MOV), try video conversion
"Processing failed"	Crash during analysis	Verify file size, disk space, try different parameters
Slow processing	Long processing times	Reduce video resolution, close other apps, check system resources
EasyOCR model download fails	Text extraction unavailable	Check internet connection, retry or restart application
"Tesseract not found"	OCR warning messages	Install Tesseract or rely on EasyOCR only (optional)
No keyframes extracted	Empty results	Lower blur threshold, adjust detection parameters
GUI doesn't start	Application won't launch	Check Python version (3.7+), install missing dependencies
Blank thumbnails	Images not displaying	Verify video codec support, try different video file
High memory usage	System slowdown	Process smaller videos, close other applications
GPU out of memory	EasyOCR crashes	Use CPU-only mode, reduce batch processing

News Detection Issues

Issue	Cause	Solution
No transitions detected	Thresholds too high	Lower headline/person/scene thresholds
Too many transitions	Thresholds too low	Increase thresholds or minimum gap
Face detection not working	Poor video quality	Use scene change detection instead
Text extraction slow	Large keyframes	Resize images or use Tesseract only
OCR accuracy poor	Low contrast text	Adjust IVP enhancement parameters

Text Extraction Issues

Issue	Symptoms	Solution
No text detected	Empty OCR results	Check text contrast, try different enhancement settings
Poor OCR accuracy	Wrong text recognition	Verify text language settings, improve image quality
Slow text extraction	Long export times	Use GPU acceleration, process fewer frames
Memory errors during OCR	Application crashes	Reduce processing batch size, use CPU-only mode
JSON export fails	Missing text results	Check write permissions, ensure sufficient disk space

Performance Issues

Problem	Indicators	Solutions
Thermal throttling	Fan noise, slow processing	Enable automatic thermal management
Memory leaks	Increasing RAM usage	Restart application, process smaller batches
GPU utilization low	Slow text extraction	Install CUDA drivers, verify GPU compatibility
Disk space full	Export failures	Clean temporary files, use different output directory

Debug Information Sources

Logs Tab: Real-time processing logs and error messages
Log Files: framesift_gui.log in application directory
Console Output: Run from terminal for detailed error traces
System Monitor: Check CPU, memory, and disk usage during processing
Thermal Status: Monitor temperature and fan status in status bar

Quick Fixes

Application Won't Start

# Check Python version
python --version

# Verify dependencies
pip install -r requirements.txt

# Run with detailed logging
python framesift_gui.py

Poor Text Extraction Results

Check image quality: Ensure text is clear and high contrast
Try different thresholds: Adjust news detection sensitivity
Use both OCR engines: Don't disable Tesseract fallback
Verify text language: Ensure OCR is configured for correct language

Memory Issues

Process smaller videos: Split large files into segments
Close other applications: Free up system resources
Use CPU-only mode: Disable GPU acceleration if unstable
Restart application: Clear memory leaks between sessions

Applications & Use Cases

Media & Broadcasting

News Analysis & Monitoring

Content Tracking: Monitor news coverage and topic transitions
Broadcast Structure: Analyze program flow and segment boundaries
Media Research: Study news presentation patterns and bias
Archive Processing: Make historical footage searchable and accessible
Real-time Analysis: Process live news streams for content changes

Video Production & Post-Production

Content Editing: Identify key moments for highlight reels
Scene Detection: Automatic shot boundary detection for editing
Quality Control: Identify technical issues and content problems
Logging & Indexing: Create detailed shot lists and content catalogs

Machine Learning & AI

Dataset Creation

Training Data: Generate labeled datasets for computer vision models
Frame Sampling: Intelligent frame selection for ML training
Text Recognition: Create OCR training datasets from video content
Face Detection: Build face recognition datasets from video sources

Content Understanding

Video Analysis: Automated content classification and tagging
Text Extraction: Extract text content for search and indexing
Object Detection: Prepare data for object recognition systems
Analytics: Quantitative analysis of video content patterns

Research & Education

Academic Research

Computer Vision Studies: Baseline implementation for keyframe extraction research
Algorithm Development: Compare and develop new IVP techniques
Performance Analysis: Benchmark different processing approaches
Media Studies: Analyze video content structure and presentation

Educational Applications

Teaching Tool: Demonstrate IVP concepts and algorithms
Hands-on Learning: Interactive parameter tuning and result visualization
Algorithm Understanding: Visual representation of processing steps
Experimentation: Safe environment for testing different approaches

Professional & Commercial

Content Management

Video Libraries: Organize and catalog large video collections
Search & Discovery: Make video content searchable by extracted text
Storage Optimization: Reduce storage requirements while preserving key information
Workflow Automation: Automate repetitive video processing tasks

Analysis & Intelligence

Performance Metrics: Quantify video content characteristics
Content Verification: Verify content accuracy and completeness
Usage Analytics: Track video engagement and important segments
Compliance Monitoring: Ensure content meets regulatory requirements

Technical Applications

System Integration

API Development: Core algorithms for larger video processing systems
Pipeline Integration: Component in automated video processing workflows
Cloud Processing: Scalable video analysis in cloud environments
Mobile Applications: Lightweight processing for mobile video apps

Development & Testing

Algorithm Prototyping: Test new video processing ideas
Performance Benchmarking: Compare processing approaches and optimizations
Quality Assurance: Validate video processing pipeline correctness
Metrics Collection: Gather performance data for system optimization

Specialized Use Cases

News & Media Monitoring

YouTube News Clips → FrameSift Analysis → Keyframe Extraction → Text Extraction → Searchable Database

Educational Content Processing

Lecture Videos → Scene Detection → Key Moment Extraction → OCR Text → Study Materials

Surveillance & Security

Security Footage → Motion Detection → Event Identification → Text Recognition → Incident Reports

Content Creation

Raw Footage → Intelligent Editing → Highlight Detection → Text Overlay → Final Production

Performance Benefits

Application Area	Time Savings	Quality Improvement	Cost Reduction
News Analysis	70-90% faster than manual review	Consistent detection accuracy	Reduced analyst workload
Video Editing	60-80% faster shot identification	Automated quality filtering	Lower post-production costs
Dataset Creation	80-95% faster than manual labeling	Systematic frame selection	Reduced annotation costs
Content Management	90%+ faster than manual cataloging	Comprehensive text extraction	Minimal storage overhead

Contributing

Development Setup

Fork & Clone: Fork the repository and clone locally
Environment Setup: Create virtual environment and install dependencies
Feature Branch: Create feature branch: git checkout -b feature/amazing-feature
Development: Make changes focusing on desktop application functionality
Testing: Run test suite and validate with demo scripts
Documentation: Update relevant documentation files
Pull Request: Submit PR with detailed description of changes

Code Style & Standards

PEP 8: Follow Python style guidelines
Type Hints: Use type annotations for function parameters and returns
Docstrings: Include comprehensive docstrings for all functions and classes
Error Handling: Implement proper exception handling and logging
Performance: Consider memory usage and processing efficiency
Cross-platform: Ensure compatibility with Windows, macOS, and Linux

Testing Guidelines

# Run all tests
python -m pytest tests/ -v

# Test specific modules
python -m pytest tests/test_news_detection.py -v

# Run integration tests
python test_selective_extraction.py
python demo_news_text_extraction.py

Areas for Contribution

Algorithm Improvements

New IVP Techniques: Implement additional image processing methods
Detection Accuracy: Improve news content detection algorithms
OCR Enhancement: Better text preprocessing for improved recognition
Performance Optimization: Faster processing without quality loss

Feature Enhancements

UI/UX Improvements: Better user interface and experience
Advanced Analytics: More detailed statistics and visualizations
Export Options: Additional output formats and integration options
Language Support: Multi-language OCR and interface support

Platform Support

Cross-platform: Improve macOS and Linux compatibility
Cloud Integration: Add cloud processing capabilities
Mobile Support: Lightweight mobile processing options
Containerization: Docker support for deployment

Conclusion

FrameSift represents a comprehensive solution for intelligent video keyframe extraction, combining traditional computer vision techniques with cutting-edge news content analysis and text extraction capabilities.

What Makes FrameSift Special

Specialized News Detection: Purpose-built for analyzing news broadcasts with ROI-based analysis
Intelligent Text Extraction: IVP-enhanced OCR with selective processing for optimal performance
Academic Foundation: Implements comprehensive IVP techniques from computer vision curriculum
Performance Optimized: Smart processing that balances speed and accuracy
User-Friendly: Desktop GUI application with intuitive workflow

Perfect For

Media Professionals analyzing news content and broadcasts
Students & Researchers studying computer vision and IVP techniques
AI Developers building video analysis and content understanding systems
Data Scientists creating training datasets from video content
Organizations processing video archives and content libraries

Results You Can Expect

Metric	Traditional Methods	FrameSift
Processing Speed	1x real-time	3-6x real-time
Frame Reduction	Manual selection	70-80% automated
Text Extraction	Manual transcription	Automated OCR with 85%+ accuracy
News Analysis	Hours of manual review	Minutes of automated processing
Workflow Efficiency	Multiple tools required	All-in-one solution

Get Started Today

Clone the repository
Install dependencies with pip install -r requirements.txt
Launch with python framesift_gui.py
Upload your first video and explore the capabilities
Try news detection on broadcast content
Export keyframes with automatic text extraction

Transform your video analysis workflow with FrameSift - where intelligent processing meets practical application!

Name		Name	Last commit message	Last commit date
Latest commit History 7 Commits
demo		demo
docs		docs
examples		examples
modules		modules
test_selective_extraction		test_selective_extraction
tests		tests
text_extraction		text_extraction
.gitignore		.gitignore
README.md		README.md
create_example_videos.py		create_example_videos.py
framesift_gui.py		framesift_gui.py
install.py		install.py
launch_gui.bat		launch_gui.bat
launch_gui.py		launch_gui.py
requirements.txt		requirements.txt

Rush-Code10/Keyframe-Detection

Folders and files

Latest commit

History

Repository files navigation

FrameSift - Advanced Keyframe Detection with News Analysis & Text Extraction

Overview

New Features (2025 Update)

Key Features

News Content Detection

Advanced Text Extraction (IVP-Based)

IVP (Image & Video Processing) Techniques

Desktop Application Features

IVP Algorithms & Techniques

Core Video Processing Algorithms

IVP Text Enhancement Pipeline

Multi-Modal Processing Pipeline

Standard Keyframe Extraction

News Content Detection

Text Extraction (Selective)

Installation

Quick Setup

Automated Installation

System Requirements

Usage

Launch Application

Application Workflow

Standard Keyframe Extraction

News Content Analysis (Specialized)

Text Extraction Features

News Content Detection & Text Extraction

Overview

What It Detects

Technical Implementation

ROI-Based Analysis

Multi-Modal Fusion

Text Extraction Pipeline

When Text Extraction Happens

IVP-Based Text Enhancement

OCR Engine Fusion

Text Region Classification

Configuration Options

News Detection Thresholds

Text Extraction Settings

Workflow Example

Step 1: Fast News Detection

Step 2: Selective Text Extraction

Output Formats

Keyframe Export

Text Extraction JSON

Export Summary Report

Use Cases & Applications

Professional Applications

Technical Applications

Parameter Configuration

Algorithm Thresholds

Available Presets

Example Results

Export Options

Project Structure

Testing

Unit Tests

Command Line Demo

Dependencies

Core Computer Vision & Processing

Text Extraction & OCR

System Management

GUI & Standard Libraries

Development & Testing Tools

Installation Commands

External Dependencies (Auto-Downloaded)

Performance & System Requirements

Processing Speed

Standard Keyframe Extraction

News Detection (Without Text Extraction)

Text Extraction (Selective Processing)

Memory Usage

Supported Formats

Input Video Formats

Output Formats

Packages