Skip to main content
All CollectionsMethodology & RequirementsRequirements
Requirements for Traffic Camera Video Processing
Requirements for Traffic Camera Video Processing

How to make your video suitable for processing from fixed traffic cameras?

Pavel Severa avatar
Written by Pavel Severa
Updated over 3 months ago

GoodVision Video Insights provides greater than 95% accuracy of collected traffic data. Our processing supports several types of vehicles, motorcycles, bicycles and pedestrians – all can be on the scene together. All you have to do is to provide a quality video input. So how to make sure that your videos have sufficient quality to achieve the best results? There are certain aspects of the video files which affect the accuracy of video analytics.

Recommended Video File Settings

Parameter

Recommended settings

Viewport

Static

Resolution

1920x1080 px (min: 640x480 px, max: 4096x2160 px)

Object Detection Size

5-30% of the screen width

Frame Rate (FPS)

25-30 FPS

Codec

H.264 or H.265

Video Length

> 1 minute

File Size

Max: 25GB

How to achieve the best results

To achieve the best results, the objects in the video must be clearly visible, without occlusions and with minimal artifacts or blurring, as our AI engine relies on computer vision to accurately detect and analyze them. If you can't see the car in the video clearly with your eyes, our AI engine will likely have trouble too.

Also make sure the camera doesn’t switch between multiple configurations and is stable during the while time of recording. Turn off night mode, avoid changing settings in the middle of the recording and turn off the option for switching to backup streams if possible.

The maximum video file size is currently limited to 25 GB. If your file exceeds this limit, we recommend using our desktop app and enabling the video optimization feature, which can significantly reduce the file size without compromising quality.

1. Camera height

The height of the camera significantly affects the quality of video analytics. Follow these guidelines for ideal camera placement:

  • Optimal Height: Position the camera at a height of 8-12 meters (26-40 feet) above the ground to capture both near and distant objects without obstruction.

  • Minimum Height: The camera should be placed no lower than 5 meters (16 feet) to prevent objects in the foreground from blocking those in the background.

  • Maximum Height: Do not place the camera higher than 30 meters (98 feet), as detection accuracy decreases at greater heights, especially for smaller objects such as pedestrians, cyclists, and motorcyclists.

  • Placement Considerations: Choose a location that allows for stable installation at the recommended height while covering the desired area. Consider the type of traffic (e.g., vehicles or pedestrians) and potential mounting points (e.g., poles, buildings).

2. Distance from monitored objects

The distance between the camera and the monitored objects is critical for reliable detection and tracking:

  • Object Size: Ensure that vehicles or objects occupy at least 5% of the total scene size. For example, in a Full HD (1920x1080 px) video, an object should have a width of at least 96 pixels.

  • Scene Coverage: Do not allow objects to cover more than 30% of the total scene size. In Full HD, an object's width should not exceed 576 pixels.

  • Camera Angle: The AI engine can process video effectively from a wide range of angles, from almost horizontal to a straight-down bird’s-eye view. Ensure that the angle does not exceed a viewing distance of 70 meters (230 feet) for all vehicle classes to be accurately tracked. Distances greater than this can cause smaller objects like pedestrians and cyclists to become less detectable.

3. Obstacles

Objects or structures in the camera's field of view can affect video analytics:

  • Small Obstacles: Elements like thin poles, standard traffic signs, and wires do not usually impact detection capabilities.

  • Large Obstacles: Larger objects such as trees, statues, buildings, and bridges can obscure objects in the scene, causing the system to misinterpret vehicle trajectories. When a vehicle reappears after being blocked, it may be considered a new object, leading to inaccurate data.

  • Placement Strategy: Position the camera to ensure that the primary traffic streams are fully visible and unobstructed. If unavoidable, consider adjusting the analysis parameters around the obstacle using tools like Smart Corridor to improve detection accuracy.

4. Resolution

Higher resolution improves the accuracy of object detection and tracking:

  • Recommended Resolutions: Use a resolution of at least 1280x720 px (HD) or 1920x1080 px (Full HD) for the best results. Resolutions higher than 1920x1080 px are also supported, up to 4096x2160 px (4K).

  • Minimum Resolution: The minimum supported resolution is 640x480 px, but using such low resolutions may cause objects to appear blurry or lack sufficient detail for accurate detection.

  • Consistent Resolution: All frames within a single video or data source must maintain the same resolution. Changes in resolution are not supported and will result in errors during processing.

Examples of blurry traffic objects on a low resolution scene

5. Frame rate (FPS)

The frame rate impacts the fluidity of object motion and tracking accuracy:

  • Ideal FPS: Set the frame rate between 25 and 30 frames per second (FPS) for the best results. This range offers a balance between smooth motion and manageable file size.

  • Maximum FPS: Frame rates above 60 FPS do not significantly improve detection accuracy but can increase file size and processing time.

  • Minimum FPS: While the minimum supported frame rate is 10 FPS, this is not recommended, especially for scenes with fast-moving objects or heavy traffic, as it can cause tracking errors.

6. Shutter speed

Shutter speed influences the clarity of moving objects, especially in low-light conditions:

  • Avoid Long Shutter Speeds: Cameras often switch to longer shutter speeds in low-light conditions, which can result in motion blur and unclear object contours. Disable automatic adjustments to maintain object clarity.

  • Maintain Clarity: Ensure the camera's settings preserve the contours of moving objects, as GoodVision Video Insights is optimized for detecting well-defined shapes, even in low-light scenarios.

7. Scene lighting

Lighting conditions directly affect the quality of video analytics:

  • Illumination Requirements: The AI engine is trained to recognize objects in both daylight and nighttime settings. However, objects must be sufficiently illuminated to be visible to the human eye.

  • Artificial Lighting: If natural lighting is inadequate, use artificial lighting to ensure that objects are visible throughout the recording period.

8. Weather and other conditions

Certain weather and environmental conditions can negatively impact video quality:

  • Rain and Snow: Avoid recording in heavy rain or snow, as raindrops or snowflakes on the lens can distort the image or cause reflections.

  • Lens Cleanliness: Ensure the camera lens is clean, free from dirt, smudges, or scratches, to avoid blurry images and preserve object contours.

  • Sun Flare and Reflections: Position the camera to minimize flare or reflections caused by direct sunlight, which can reduce object clarity.

  • Avoid Barrel Distortion: Choose a camera and lens setup that minimizes distortion, which can cause objects to appear bent or warped.

By adhering to these detailed requirements, you will ensure that your video footage is optimal for processing with our AI engine, leading to accurate and reliable results.

Did this answer your question?