No-reference image quality assessment for iterative batch video analysis
AssigneeArlo Technologies, Inc.
InventorsPeiman Amini, Joseph Amalan Arul Emmanuel
Life of this patent
Estimating the term…
Abstract
Introduced here are technologies for examining content generated by electronic devices in real time to optimize the quality of the content. The content may be examined in batches to address some of the drawbacks of real-time analysis. For instance, a series of videos may be examined to collect data on how well security system(s) that are presently employed are working. Each security system can include one or more electronic devices, such as cameras or microphones, and parameters of the electronic devices can be altered to improve the quality of content generated by the electronic devices.
Claims20
A computer-implemented method comprising:
- receiving, by a computer system, a first video captured by a camera having a parameter set to a first value based on pre-launch training of a reference-less machine learning model using a database of videos having known distortions;
- extracting a feature vector from the first video, wherein the feature vector describes distortions present in the first video;
- determining, using the reference-less machine learning model, a quality metric of the first video based on the feature vector,
- wherein the quality metric indicates statistical properties of the first video corresponding to the distortions;
- determining a second value for the parameter of the camera based on the quality metric of the first video;
- and
- causing the camera to set the parameter to the second value,
- wherein the quality metric of a second video captured by the camera is greater than the quality metric of the first video when the parameter is set to the second value.
An electronic device comprising:
- a network interface for communicating with a camera over a wireless communication channel;
- a processor;
- and
- a memory having instructions stored thereon that, when executed by the processor, cause the electronic device to:
- receive a first video captured by the camera when the camera has a parameter set to a first value based on pre-launch training of a reference-less machine learning model using pixel data or parametric data in a database of videos;
- extract a feature vector from the first video;
- determine, using the reference-less machine learning model, a quality metric of the first video based on the feature vector,
- wherein the quality metric indicates statistical properties of the first video;
- determine a second value for the parameter of the camera based on the quality metric of the first video;
- and
- cause the camera to set the parameter to the second value,
- wherein the quality metric of a second video captured by the camera increases when the parameter is set to the second value.
A computer system comprising:
- a processor;
- and
- a memory having instructions stored thereon that, when executed by the processor, cause the computer system to:
- receive video captured by a camera when the camera has a parameter set to a first value based on pre-launch training of a reference-less machine learning model using pixel data or parametric data in a database of videos;
- determine, using the reference-less machine learning model, a quality metric of the video;
- cause the camera to set the parameter to a second value determined based on the quality metric of the video;
- detect motion in the video;
- and
- in response to detecting motion in the video, cause a security light communicably coupled to the computer system to:
- illuminate an environment depicted by the video.
Dependencies are inferred from the claim text.