Motion Detection types (Frame based vs Experimental)

Overview

Redcherry provides three motion detection algorithms to suit different surveillance environments and requirements. Each algorithm offers different levels of accuracy, CPU usage, and configuration options.

Motion Detection Algorithms

🟢 DEFAULT (Original Motion Detection)

Best for: Simple, stable environments with consistent lighting and limited CPU resources.

Characteristics:

• Fast and lightweight processing
• Lower CPU usage
• Immediate motion response
• No configurable sliders (uses default settings)
• Ideal for basic surveillance with just a few cameras

How it works: Uses a single reference frame that slowly adapts to scene changes, comparing each new frame against this background reference.

🔵 EXPERIMENTAL (Basic Motion Detection)

Best for: Simple, stable environments with consistent lighting when you need quick response to movement.

Characteristics:

• Compares each video frame against two reference frames
• One reference updates quickly, the other more slowly
• Triggers motion when changes exceed defined thresholds
• Lower CPU usage than advanced mode
• Immediate motion response

Available Controls:

• Frame downscale factor
• Min. motion area %
• Max. motion area %

🟡 FRAME BASED (Advanced Motion Detection)

Best for: Environments with changing lighting, background motion (trees, shadows, bugs), or when motion detection accuracy is critical.

Characteristics:

• Uses sophisticated temporal analysis across multiple frames
• Continuously updated reference frame with motion event history
• Higher accuracy with fewer false positives
• Adapts better to dynamic environments
• Requires more CPU and memory, but suitable for any modern PC

Available Controls:

• All sliders enabled
• Temporal frame buffer analysis
• Reference frame blending

Configuration Parameters

Frame Downscale Factor

Purpose: Reduces frame resolution before motion analysis to improve performance.

Options: 0.125, 0.25, 0.5, 1.0

How it works: A value of 0.5 means the frame is scaled down to 50% of original size before processing.

Impact:

• Lower values (0.125-0.25): Faster processing, less detail, suitable for high-resolution cameras
• Higher values (0.5-1.0): More detail, slower processing, better for low-resolution cameras

Recommendation: Start with 0.5 for most cameras. Use 0.25 for 4K cameras or systems with limited CPU.

Miniminal Motion Area %

Purpose: Sets the minimum percentage of the frame that must show motion to trigger detection.

Range: 1-100%

How it works: The system calculates what percentage of the total frame area contains motion. Only triggers if motion covers at least this percentage.

Practical Examples:

• 1-5%: Detects small objects (people, animals) moving in the distance
• 5-15%: Good for detecting people walking through the frame (recommended starting point)
• 15-30%: Detects larger movements, reduces false positives from small changes
• 30%+: Only triggers on large objects or multiple people

Recommendation: Start with 5-15% for most surveillance scenarios.

Maximum Motion Area %

Purpose: Sets the maximum percentage of frame motion before ignoring it as “too much motion.”

Range: 1-100%

How it works: If motion covers more than this percentage, it’s considered noise (camera shake, lighting changes) and ignored.

Practical Examples:

• 50-70%: Allows moderate motion before ignoring
• 70-90%: Good for stable cameras with occasional large movements
• 90%+: Very permissive, only ignores extreme motion

Recommendation: Set to 70-90% for most applications.

Miniminal Motion Frames (FRAME BASED only)

Purpose: Sets how many consecutive frames must show motion before triggering an event.

Range: 1-255

How it works: Uses a temporal buffer to track motion across multiple frames. Only triggers when motion is detected in at least this many frames within the buffer.

Impact:

• Lower values (1-5): Fast response, more false positives
• Higher values (10-20): More stable detection, fewer false positives

Recommendation: Start with 10-15 frames for most environments.

Maximum Motion Frames (FRAME BASED only)

Purpose: Sets the size of the frame buffer used for temporal analysis.

Range: 1-255

How it works: This is the “ring buffer” size that stores the motion history of recent frames for temporal analysis.

Impact:

• Smaller buffer (10-20): Faster response, less temporal context
• Larger buffer (20-50): Better temporal analysis, more memory usage

Recommendation: Set to 20-30 frames for most applications.

Reference Frame Blending Ratio (FRAME BASED only)

Purpose: Controls how quickly the reference frame adapts to changes in the scene.

Range: 2-100

How it works: The reference frame is continuously updated by blending new frames with the existing reference. This ratio determines the blend percentage.

Impact:

• Lower values (2-20): Slow adaptation, better for stable environments
• Higher values (20-50): Fast adaptation, better for dynamic environments

Recommendation: Start with 10-20 for most environments.

Motion Sensitivity Grid

The motion sensitivity grid provides preset sensitivity levels that affect the motion detection thresholds:

• Off: Motion detection disabled
• Minimal: Only detects large, significant movements
• Low: Detects moderate to large movements
• Average: Balanced sensitivity for most scenarios (recommended starting point)
• High: Detects smaller movements and subtle changes
• Very High: Maximum sensitivity, detects even very small movements

Note: These sensitivity levels are about motion detection threshold sensitivity, not object size. High sensitivity detects smaller movements, while low sensitivity only detects larger movements.

Configuration Recommendations

For Indoor Surveillance

• Algorithm: BASIC or FRAME BASED
• Min. Motion Area: 5-10%
• Max. Motion Area: 70-80%
• Frame Downscale: 0.5
• Sensitivity: Average to High

For Outdoor Surveillance

• Algorithm: FRAME BASED
• Min. Motion Area: 10-20%
• Max. Motion Area: 80-90%
• Frame Downscale: 0.25-0.5
• Min. Motion Frames: 15-20
• Sensitivity: Low to Average

For High-Traffic Areas

• Algorithm: FRAME BASED
• Min. Motion Area: 15-25%
• Max. Motion Area: 85-95%
• Min. Motion Frames: 20-25
• Sensitivity: Low

For Low-Light Environments

• Algorithm: FRAME BASED
• Min. Motion Area: 8-15%
• Max. Motion Area: 75-85%
• Reference Frame Blending: 15-25
• Sensitivity: Average

Troubleshooting

Too Many False Positives

• Increase Min. Motion Area %
• Increase Min. Motion Frames (FRAME BASED)
• Lower sensitivity level
• Increase Max. Motion Area %

Missing Important Events

• Decrease Min. Motion Area %
• Decrease Min. Motion Frames (FRAME BASED)
• Increase sensitivity level
• Switch to FRAME BASED algorithm

High CPU Usage

• Decrease Frame Downscale Factor
• Switch to CV_BASIC algorithm
• Reduce Max. Motion Frames (FRAME BASED)

Poor Performance in Changing Lighting

• Switch to FRAME BASED algorithm
• Adjust Reference Frame Blending Ratio
• Increase Min. Motion Frames

Debug Features

Enable Debugging Snapshots

When enabled, the system saves motion detection debug images to help troubleshoot configuration issues. These images show:

• Motion detection areas
• Reference frame comparisons
• Threshold analysis

Location: Debug images are typically saved to /tmp/ or the system’s temporary directory.

Best Practices

1. Start with defaults and adjust based on your specific environment
2. Test thoroughly in your actual surveillance environment
3. Monitor CPU usage and adjust Frame Downscale Factor if needed
4. Use FRAME BASED for environments with changing lighting or background motion
5. Enable debug snapshots temporarily when fine-tuning settings
6. Consider camera placement - avoid pointing at moving objects like trees or flags
7. Regular maintenance - review and adjust settings as environmental conditions change