US 12,467,818 B2
Detecting gas leaks from image data and leak detection models
Anatoly Aseev, Menlo Park, CA (US); Andrey Sergeevich Konchenko, Menlo Park, CA (US); Jose R. Celaya Galvan, Sunnyvale, CA (US); Indranil Roychoudhury, Sunnyvale, CA (US); and Prasham Sheth, Sunnyvale, CA (US)
Assigned to Schlumberger Technology Corporation, Sugar Land, TX (US)
Filed by Schlumberger Technology Corporation, Sugar Land, TX (US)
Filed on Dec. 13, 2023, as Appl. No. 18/537,895.
Claims priority of provisional application 63/477,590, filed on Dec. 29, 2022.
Prior Publication US 2024/0219255 A1, Jul. 4, 2024
Int. Cl. G01M 3/04 (2006.01); G05B 15/02 (2006.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/77 (2022.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 20/52 (2022.01)
CPC G01M 3/04 (2013.01) [G05B 15/02 (2013.01); G06V 10/44 (2022.01); G06V 10/764 (2022.01); G06V 10/7715 (2022.01); G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 20/52 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
receiving, via one or more processors, a first set of image data representative of equipment configured to distribute a gas;
extracting, via the one or more processors, one or more features from the first set of image data, wherein the one or more features are representative of one or more gas leaks associated with the equipment, wherein the one or more features are identified by applying a mask regional convolutional neural network to the first set of image data to determine one or more candidate object bounding boxes associated with the one or more gas leaks;
generating, via the one or more processors, a leak detection model based on the one or more features, wherein the leak detection model correlates the extracted features of the one or more gas leaks with a corresponding calibrated output of the leak detection model, wherein the leak detection model provides a weight for each correlation corresponding to a strength of the correlation;
receiving a second set of image data;
determining, via the one or more processors, a type of the equipment depicted in the second set of image data;
retrieving, via the one or more processors, the leak detection model corresponding to the type of the equipment depicted in the second set of image data;
training, via the one or more processors, the leak detection model to accurately detect the one or more gas leaks based on the second set of image data provided to the leak detection model, wherein the second set of image data is associated with a test image;
in response to determining, via the one or more processors, that the trained leak detection model accurately detects leaks based on the second set of image data;
determining, via the one or more processors, that a gas leak is present on the equipment based on the second set of image data and the trained leak detection model;
sending, via the one or more processors, a notification to a computing device in response to detecting the one or more gas leaks; and
sending, via the one or more processors, a command to a device to cause the device to close to prevent gas from being supplied to the one or more gas leaks.