Technology Review — Computer Vision and How Can We Use It in The Energy Sector?

Written by Kristjan Eljand | Technology Scout

Computer Vision is a branch of artificial intelligence which aims to extract information from digital images and videos. The most common functionalities of Computer Vision are:

• Object detection and classification;
• Text recognition from images (identification of both digital and handwritten text is possible);
• Recognition of faces and emotions.

1. Object detection

The following figure (Figure 1) shows an example of using Computer Vision for identifying and classifying objects. In this case, the model successfully realized that there is Power Line & Electrical Supply on the image and that it is a Public utility. The model was also able to define the location of the object on the image.

Figure 1. Example of using Computer Vision to identify objects

Computer Vision enables to identify objects (what is on the picture), make classifications (broken/not broken; big/small) and find the object’s location on pictures (3cm from the left, 2cm from the top).

Potential use cases of object detection in the Energy industry:

• IT Solution for wind energy producers — regular drone images are used to identify possible maintenance needs: dirt accumulated on wind turbines, dead insects, salt (offshore wind turbines); cracks in foundation or tower;
• Identifying the best solar power production sites from aerial photographs and satellite imagery.
• IT solution for mines — automatic detection of anomalies in mines (leaks, fires, smoke, etc.) or production blocks;
• IT solution for production — cameras installed in a production facility regularly take digital images, which carry out automatic machine inspection and detect anomalies.
• IT solution for transport equipment managers — a digital photo is regularly taken of transport equipment to identify any broken parts.
• IT Solution for network service providers — Regular drone images automatically detect the need for network maintenance.

2. Text recognition

Figure 2 shows an example of using the Computer Vision for extracting text from digital images. This functionality is also called Optical Character Recognition (OCR). It is worth noting that today’s Computer Vision technologies are also capable of extracting text from the handwritten text.

Figure 2. An explanation of text recognition from Estonia mine´s label

Computer Vision can be used to extract text from images, pdf files, and non-machine-readable (archived documents) documents.

Text recognition use cases in the Energy industry:

• Digitizing handwritten documents — OCR is used to digitize handwritten documents.
• Automating pdf-related tasks — automatic search and information gathering from pdf documents (example: product BOM (Bill-of-Materials) is automatically recognized and saved to the digital database).
• Automating contract management — specific parts of contracts (for example: potential monetary liabilities) are searched and highlighted automatically.

3. Recognition of faces and emotions

In the third example (below Figure 3), Computer Vision has been used to identify faces and emotions on images. In this case, we can see that the faces of all three people have been successfully identified and the average person in the picture is likely to be happy (Joy: Likely).

Figure 3. Example of using Computer Vision to recognize faces and emotions

Face recognition can be used to identify people and read their emotions.

Face and emotion recognition use cases in the Energy sector:

• IT solution for mines — automatic identification of workers entering the mine;
• The computer’s web camera monitors the user’s eye movement and emotions in the energy provider’s self-service portal, an algorithm calculates the most relevant new services for the customer and bottlenecks in the self-service portal.

All in all, Computer Vision is one of the most advanced branches of artificial intelligence today, but it’s just starting to emerge in the workflows of organizations. There is a lot of potential, however, because Computer Vision can potentially automate all the tasks where vision is the most important sense in terms of carrying out the task.