ViFoSe (Video Foreground Segmentation)
ViFoSe is a MATLAB application specialized in foreground segmentation of video sequences, developed at the University of Bologna. The platform integrates manual and automatic segmentation tools for the analysis and processing of video content in academic and professional settings.
- INSTALLATION AND STARTUP -System Prerequisites: MATLAB R2020b or later Image Processing Toolbox 8 GB RAM (16 GB recommended for high-resolution video)
-Startup Procedure: Start the MATLAB environment Execute the command: app = ViFoSe_1; The graphical interface will initialize automatically
- INTERFACE ARCHITECTURE -Display Panels: UIAxes (Left): Display of the preprocessed video UIAxes2 (Center): Representation of the generated binary masks UIAxes3 (Right): Preview of the output video with applied masks
-Command Structure: 🎥 DATA ACQUISITION MODULE Load Video: Load video file (supported formats: MP4, AVI, MOV) Load Project: Reconstruct a work session from a saved project
✂️ PREPROCESSING TOOLS Crop ROI: Define a region of interest via rectangular selection Frame Navigation: Inter-frame navigation controls (+/-, advanced slider)
🎨 MANUAL SEGMENTATION Segment Foreground: Annotate foreground areas (green color) Segment Background: Annotate background areas (red color) Reset Segmentation: Selective reset of annotations
🤖 AUTOMATIC SEGMENTATION Automatic Segmentation: Integration with the SAM framework Automatic Alignment: Alignment via the STABA algorithm Manual Alignment: Manual alignment with STABM Load Automatic Binary Masks: Import pre-computed masks
💾 OUTPUT MANAGEMENT Save Project: Export complete project About Us: Development contact information
- STANDARD OPERATING PROCEDURE 3.1 Initialization Phase: Select "Load Video" Choose the source video file Wait for the frame extraction to complete (progress indicator) Verify the correct display of the first frame in the three panels
3.2 Region of Interest Definition (Optional): Click "Crop ROI" Delimit the rectangular area of interest Confirm the selection The system will automatically apply the crop to all frames Existing masks will be reconfigured to the new geometry
3.3 Manual Segmentation: -Foreground Procedure: Select "Segment Foreground" Draw polylines on the objects of interest (green annotation) Specify the temporal application interval Confirm for multi-frame application
-Background Procedure: Select "Segment Background" Delineate background areas (red annotation) Define the target frame interval Apply the segmentation
3.4 Validation and Navigation: Use the "Frame +/-" controls for sequential navigation Use the slider for random access to frames Monitor in real-time the effect of the masks in the three views
3.5 Process Automation: -SAM Automatic Segmentation: Activate "Automatic Segmentation" Follow the SAM protocol in the dedicated interface The generated masks will be imported automatically
-Loading External Masks: Select "Load Automatic Binary Masks" Locate the directory containing the binary masks Specify the reference time interval Masks will be imported as foreground layers
3.6 Exporting Results: Execute "Save Project" Select the destination directory The system will generate: [name]_preprocessedVideo.avi - Preprocessed video [name]_binaryMask.avi - Binary masks [name]_outputVideo.avi - Final composition
- ADVANCED FEATURES -Multiple Mask Management: Support for multiple masks per frame Temporal hierarchy: recent masks overwrite previous ones Conflict management via chronological ordering
-Selective Reset: Targeted deletion of masks in specific intervals Preservation of work unaffected by changes
-Alignment Systems: STABA: Automatic alignment for videos with camera motion STABM: Manual correction of alignments
- TROUBLESHOOTING -Error Codes: "No video loaded": A video must be loaded before processing "Mask dimensions do not match": Mask size misalignment Invalid ROI: Repeat the region selection procedure
-Operational Guidelines: Define an initial ROI for performance optimization Incremental project saving Multi-frame verification before applying masks to large intervals
- TECHNICAL SUPPORT For technical assistance, bug reports, or feature requests: Use the "About Us" panel Contact the development team at the University of Bologna
ViFoSe - Video Foreground Segmentation Tool Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi" University of Bologna, Italy Copyright © 2026 - All rights reserved
