Video Parameters
The following are some technical video parameters and the respective recommendations for effective video analytics.
Bit Rates
When using BriefCam, setting an unlimited bit rate can result in large file sizes that hinder the product's ability to fetch videos efficiently from the VMS. Therefore, it is recommended to limit the bit rate to a level that is optimal for the resolution of the video. Beyond a certain threshold, increasing the bit rate does not add any real value to the video.
The optimal bit rate for 1080p video can vary depending on a variety of factors, such as the complexity of the scene, the amount of motion, and the desired level of quality. However, as a general guideline, it is recommended to use a bit rate between 2,500 and 5,000 kbps for 1080p video. This bit rate range should provide a good balance between video quality and file size.
Framerates
A framerate is how often a camera produces frames and is usually measured in FPS or frames per second. In general, when possible, always set the video input to a constant framerate.
Windows-based OX5 Engine
For the Windows-based OX5 engine, BriefCam highly recommends a minimum frame rate of 8 frames per second (FPS). A frame rate lower than 8 can be used, but the quality of the detection, tracking, and classification is affected.
The maximum recommended frame rate is up to 15 frames per second. If a camera is configured for a higher number of frames per second, BriefCam’s processing algorithm lowers the frame rate by skipping some of the frames. In the table below you can see how many frames BriefCam processes and how many frames are skipped.
FPS | Processed Frames | Details |
6 | 6 | Video is processed as is, with performance degradation (i.e. reduced event detection rate and potential misses to be expected). |
15 | 15 | Video is processed as is, at the input frame rate (i.e. each individual frame is processed, as opposed to alternating frames). |
16 | 8 (16÷2=8) | BriefCam processes alternating frames, effectively reducing the number of frames to be processed by half. |
25 | 12 (25÷2=12.5) | |
60 | 15 (60÷2=30, then 30÷2=15) |
Linux-based OX6 Engine
For the Linux-based OX6 engine, BriefCam works at a framerate of approximately 3 FPS (or lower). However, due to the low-FPS processing, high-speed objects across the frame may not be detected.
Resolution
The resolution needs to match the desired analytics goals and the camera's field of view. The minimum detection and classification resolutions need to be met, where the higher the resolution the better the analytics accuracy, particularly for face recognition where the more facial features that are seen, the better the accuracy. For example, an outdoor camera covering an entire intersection should be 4K to allow faces to be efficiently recognized.
Variable-resolution cameras that change resolutions cannot be used with BriefCam.
Different resolutions between archive and real-time multi-streams can be different – as long as they both have the same aspect ratio.
Videos that are recorded with a variable-FPS rate might cause the tracking and detection mechanisms to behave sub-optimally in cases of counting. This is because during BriefCam engine’s 5-minute warmup, the engine determines the rate of FPS and uses this number to sample frames for detection. For example, if 15 FPS is used, every 5th frame is used for detection. For variable-rate FPS, the number selected in the 5-minute warmup is probably incorrect for different parts of the movie, which leads to jittery movement as perceived by the algorithm and unpredictable detection and tracking accuracy. It is recommended to not use variable-rate FPS and to set the camera to fixed-framerate.
Face Recognition Required Resolution
The minimum resolution required for successful face recognition is 40x40 pixels across the person’s face, 20 pixels between the eyes, or 400 pixels per meter.
You can use this size to calculate the distance at which BriefCam can perform face recognition by taking into consideration the lens FOV (field of view) angle and the number of pixels in the camera sensor.
The steps for calculating the distance are:
Calculate the width (or height) in meters of the FOV at a certain distance using the horizontal – or vertical FOV angle (and the tangent function).
See how many pixels there are per this width (or height) – by checking the sensor spec.
Calculate the width (and height) in meters per pixel (by dividing the value from step 1 by the value in step 2).
The table below is based on this minimum resolution limit and approximates the maximum operational distance of BriefCam’s face recognition engine per different camera resolutions and horizontal field of view (in degrees).
Face recognition - Approximate maximum distance [m] | |||
Resolution | HFOV*=40deg | HFOV=60deg | HFOV=90deg |
720p | 6.5 m/22 feet | 4.5 m/15 feet | 3.3 m/11 feet |
1080p | 10 m/33 feet | 7 m/23 feet | 5 m/16.5 feet |
*HFOV = horizontal field of view