Facial-recognition cameras in London

This item appears on page 4 of the March 2020 issue.
This is subscriber only post.
Get one year of online-only access — only $15!
Below is a sample of the article.
Please login or subscribe to ITN to read the entire post.

The London Metropolitan Police announced on Jan. 24 that parts of the city would be outfitted with facial-recognition cameras, with rollout expected to be completed by the end of February.

According to the police, signs will be posted at all cameras indicating that they are facial-recognition cameras, and they will be operated five to six hours at a time. The software used compares the faces on camera only with faces of wanted criminals; it does not identify all individuals scanned.

An official report on trial runs suggested that the system accurately identified 70% of designated suspects. However, an independent review found that only 19% of the facial matches were verifiably correct. Additionally, facial-recognition software has been found to be less accurate when identifying people with darker skin, due to software algorithms having been built using predominantly people with white faces.

Please login or subscribe to ITN to read the entire post.

The London Metropolitan Police announced on Jan. 24 that parts of the city would be outfitted with facial-recognition cameras, with rollout expected to be completed by the end of February.

According to the police, signs will be posted at all cameras indicating that they are facial-recognition cameras, and they will be operated five to six hours at a time. The software used compares the faces on camera only with faces of wanted criminals; it does not identify all individuals scanned.

An official report on trial runs suggested that the system accurately identified 70% of designated suspects. However, an independent review found that only 19% of the facial matches were verifiably correct. Additionally, facial-recognition software has been found to be less accurate when identifying people with darker skin, due to software algorithms having been built using predominantly people with white faces.