Video Logo Scanner

A Python script that scans videos in a folder to detect a specific logo (like the ooTruffle logo) and allows bulk moving of matching files.

Features

• Scans all videos in a folder for a specific logo
• Corner-specific detection: Only searches in corner regions where logos typically appear (faster and more accurate)
• Uses template matching to detect the logo in video frames
• Saves results to JSON for easy management
• Bulk move matching videos to a destination folder
• Configurable matching threshold, frame sampling, and corner regions

Installation

1. Install Python 3.7 or higher
2. Install dependencies:

pip install -r requirements.txt

Usage

Step 1: Prepare Template Image(s)

First, you need a screenshot or image of the logo you want to detect. You can:

• Take a screenshot from a video that contains the logo
• Extract a frame from a video using the script
• Use any image file containing the logo

For better accuracy (text + icon detection): If your logo has multiple elements (like text + an icon), you can use dual-template matching:

1. Use your full logo as the main template
2. Extract just the icon portion (e.g., shovel) as a second template using extract_shovel_template.py
3. Use --require-both to ensure both elements are present

Step 2: Scan Videos

Scan all videos in a folder:

python video_logo_scanner.py scan --folder ./videos --template logo.png

Options:

• --folder: Folder containing videos to scan
• --template: Path to the logo template image (required)
• --threshold: Matching threshold (0.0-1.0, default: 0.7). Higher = stricter matching
• --sample-frames: Number of frames to sample per video (default: 10)
• --output: Output JSON file for results (default: scan_results.json)
• --quiet: Suppress progress output
• --corners: Which corners to check: top-left, top-right, bottom-left, bottom-right (default: all corners)
• --corner-size: Fraction of frame size for corner region (0.0-1.0, default: 0.2 = 20%)
• --second-template: Path to a second template image (e.g., just the shovel icon)
• --require-both: Require BOTH templates to be found (prevents false positives from text-only matches)

Examples:

# Scan only top-right corner (where logo typically appears in kill feed)
python video_logo_scanner.py scan --folder ./videos --template logo.png --corners top-right

# Require both text and shovel icon (prevents false positives)
python video_logo_scanner.py scan --folder ./videos --template logo.png --second-template shovel.png --require-both

# Scan with custom threshold and corner size
python video_logo_scanner.py scan --folder ./videos --template logo.png --threshold 0.8 --sample-frames 20 --corner-size 0.25

# Scan multiple specific corners
python video_logo_scanner.py scan --folder ./videos --template logo.png --corners top-right top-left

Step 3: Move Matching Videos

After scanning, move all matching videos:

python video_logo_scanner.py move --results scan_results.json --destination ./matched_videos

Or move specific files:

python video_logo_scanner.py move --files video1.mp4 video2.mp4 --destination ./matched_videos

Options:

• --results: JSON file with scan results
• --files: Specific files to move (space-separated)
• --destination: Destination folder (required)
• --preserve-structure: Preserve directory structure when moving

How It Works

1. Corner Region Extraction: Only searches in specified corner regions (e.g., top-right for kill feed logos)
2. Template Matching: Uses OpenCV's template matching to find the logo in corner regions
3. Frame Sampling: Instead of checking every frame, it samples frames evenly throughout each video
4. Confidence Scoring: Each match is scored by confidence (0.0 to 1.0)
5. Results Storage: Matching videos are saved to a JSON file with their paths and confidence scores

Why corner detection? Logos in videos (like player names in kill feeds) typically appear in fixed corner positions. By only searching corners, the script is:

• Faster: Searches ~20% of each frame instead of 100%
• More Accurate: Reduces false positives from similar patterns elsewhere in the frame
• More Reliable: Focuses on where the logo actually appears

Tips

• Better Template: Use a clear, high-quality screenshot of just the logo for best results
• Corner Selection:
  • For kill feed logos (like ooTruffle), use --corners top-right
  • If unsure, scan all corners first, then narrow down based on results
• Corner Size:
  • Default 20% works for most cases
  • Increase to 25-30% if logo is larger or positioned further from corner
  • Decrease to 15% if logo is very small and close to corner edge
• Threshold Tuning:
  • Lower threshold (0.5-0.6): More matches, but may include false positives
  • Higher threshold (0.8-0.9): Fewer matches, but more accurate
• Frame Sampling: More frames = more thorough but slower scanning
• Video Formats: Supports common formats: .mp4, .avi, .mov, .mkv, .flv, .wmv, .m4v, .webm

Example Workflow

# 1. Extract a frame from a video to create template (optional)
python extract_frame.py video_with_logo.mp4 logo_template.png --frame 100

# 2a. (Optional) Extract just the shovel icon for dual-template matching
python extract_shovel_template.py logo_template.png shovel_template.png

# 2b. Scan videos (checking only top-right corner where logo appears)
# Option A: Single template (simpler, but may match text-only)
python video_logo_scanner.py scan --folder ./my_videos --template logo_template.png --corners top-right

# Option B: Dual template (more accurate, requires both text and shovel)
python video_logo_scanner.py scan --folder ./my_videos --template logo_template.png --second-template shovel_template.png --require-both --corners top-right

# 3. Review results in scan_results.json

# 4. Move matching videos
python video_logo_scanner.py move --results scan_results.json --destination ./videos_with_logo

Troubleshooting

• No matches found: Try lowering the threshold or using a better template image
• Too many false positives:
  • Use --require-both with a second template to require both text and icon
  • Increase the threshold
  • Improve the template quality
• Finding text but not icon: Use --second-template and --require-both to require both elements
• Slow scanning: Reduce the --sample-frames value
• Template not loading: Ensure the image file exists and is in a supported format (PNG, JPG, etc.)