If you are using object detection algorithms or machine learning models to analyze images, it's essential to have high-quality images that can provide accurate data. One critical factor to consider when taking images is lighting. Proper lighting can make a significant difference in the quality of the image and ultimately improve the accuracy of the analysis.
In this article, we will discuss the importance of lighting when taking images and how to optimize your camera settings to improve object detection accuracy.
The Importance of Lighting
Lighting plays a critical role in image quality because it affects the contrast, color, and sharpness of the image. Poor lighting can result in blurry or distorted images, which can make it difficult for object detection algorithms to accurately identify objects within the feed.
Optimizing Camera Settings
There are several camera settings you can adjust to optimize the Camera's image quality. The following are the settings that have to be adjusted:
The exposure determines how much light enters the camera. Increasing the exposure can make the image brighter, but it can also result in overexposed images with blown-out highlights. Decreasing the exposure can make the image darker, but it can also lead to underexposed images with lost shadow detail. We recommend setting the exposure to auto, which adjusts the exposure based on the lighting conditions.
But when it comes to camera settings for low-light conditions, the fully automatic mode may not provide sufficient lighting, especially during the night or for indoor cameras with low levels of artificial lighting. In such cases, it's essential to use the right camera settings to improve the image quality and object detection accuracy.
Most cameras have different modes such as Automatic, Manual, and Semi-Automatic modes. For Uniview cameras, the Semi-Automatic mode is called "Low Motion Blur" and is recommended for low-light conditions. In this mode, the user can specify the slowest shutter speed allowed, which helps to capture more light and reduce motion blur.
It is recommended to set the shutter speed to 1/120 or less, provided there is enough light during the darkest time of the day. In case the camera cannot capture enough light at 1/120, the shutter speed can be decreased to 1/60, but this will introduce some motion blur.
By using the "Low Motion Blur" mode and setting the shutter speed to the appropriate level, you can capture high-quality images even in low-light conditions, which will improve object detection accuracy.
In the provided images, the exposure was adjusted manually to provide a better balance between the highlights and shadows.
In this example, in bright daylight, a fast exposure is selected else the image will be too bright in the day time. The issue with this is at night this fast exposure will make the image too dark and this will cause an underexposed image. Usually to achieve a perfect balance, set the exposure to auto but if that is causing motion blur then try adjusting the exposure to different environments like on a rainy day, in the evening, at night time, and when the sun is fully exposed and get an average setting.
If the exposure is having slow shutter value then it will make the image too bright and the system won't be able to detect objects in the image area where the light is too much.
When fast exposure is used it can cause the area to be underlit and dark objects to appear similar, in the example below the under-exposure caused the image darker in the middle and the system failed to detect 2nd person
If the image is dark and the object is not detectable, try the image adjustment option on the camera. Increasing the Contrast and Sharpness usually helps in the differentiation of objects. Try keeping the brightness in the middle and try decreasing the saturation. The lower saturation might cause the image to be less vivid but make the contrast better for object detection.
White balance adjusts the color temperature of the image to make it look natural. Different lighting conditions can produce different color casts, which can affect the accuracy of the analysis. In the provided images, the white balance was adjusted to improve color accuracy and make the image look more natural.
WDR stands for Wide Dynamic Range, and it's a feature that helps to balance the brightness and contrast of images when there is a significant difference between the brightest and darkest parts of the image. There are two types of WDR: hardware and software.
Software WDR is typically less effective than hardware WDR because it relies on software processing to balance the brightness and contrast of the image, while hardware WDR uses specialized hardware to achieve the same result. Cameras with hardware WDR can achieve much better results in poor lighting conditions with WDR enabled.
However, it's important to note that too much WDR compensation can also start to blur the image, which can reduce the effectiveness of the feature. So it's important to find the right balance between WDR compensation and image sharpness when using this feature.
Digital Noise Reduction
Digital noise reduction is a feature that is commonly found in modern cameras. It works by removing the unwanted noise that is present in the image, which can result in a cleaner and more detailed image. The two types of digital noise reduction are 2D and 3D noise reduction.
2D noise reduction is used to remove noise from the image in the horizontal and vertical directions. This type of noise reduction is effective for reducing the noise that is caused by the camera's sensor.
3D noise reduction is used to remove noise from the image in all three dimensions. This type of noise reduction is more effective than 2D noise reduction and can result in a much cleaner and more detailed image.
When adjusting the digital noise reduction settings on your camera, it is essential to find the right balance between noise reduction and image blurring. Increasing the noise reduction too much can lead to an overly smooth and blurred image, which can reduce the accuracy of the analysis. Therefore, it is recommended to increase the noise reduction gradually until the optimal balance between noise reduction and image quality is achieved.
Defog is an image enhancement feature that can improve visibility in foggy or hazy conditions. It works by enhancing contrast and reducing glare, which can help to make objects in the image more visible.
Note: If the Defog option is grayed out, try turning off the WDR feature to enable Defog. Some cameras cannot use both features simultaneously, and turning off WDR may allow you to access Defog.
To optimize your camera settings for object detection, we recommend the following:
Use natural lighting whenever possible. Avoid harsh artificial lighting or direct sunlight, which can create shadows and highlights that affect the accuracy of the analysis.
Experiment with different camera settings to find the optimal balance between exposure, and white balance.
While these features can be useful in certain situations, it's important to be careful not to overdo it. Too much image enhancement can lead to artifacts or distortions in the image, which can negatively impact object detection accuracy. It's best to experiment with these features and find a balance that works for the specific environment and camera setup. In addition, it's important to remember that these features can be computationally expensive, so it's important to consider the processing power of the camera or system when using them.
Please sign in to leave a comment.