NOAA Set To Launch Face Detection Software To Identify Right Whales
The National Oceanic and Atmospheric Administration (NOAA), USA is all set to launch a facial detection software to identify right whales. NOAA along with Mathworks put forward a competition on Kaggle, a data science venture for enthusiasts to develop a software that can rightly detect “right whales†and uniquely identify them. The competition ended on 7th January, 2016 and the data was recorded to be published as a press release.
Christin Khan, a member of NEFSC, under NOAA, works on aerial surveys that capture snapshots of whales from the top, said that identifying whales using mere photography was tedious and there was a high chance of error. To overcome the problem, the NOAA decided to use the technique of facial recognition.
The underlying principle behind the face recognition software was to uniquely categorize callosities or lice patterns. The challenge for data scientists was to automate the right whale recognition process by using given datasets of aerially photographed whales. After a 4-month period, deepsense.io, a software engineering firm overpowered other competitors and grabbed the $10,000 prize. Polish creator Robert Bogucki explained that his team’s whale algorithm would need more data points to improve its precision as the dataset only had 4500 whale photographs.
Right whale identification
Christin claimed that the software would be incorporated in their searching methods within the next two years. This new advancement in their searching technique would definitely enhance the identification process of endangered whales.
Watch Right whale "Calvin" with her calf -
Source: <a href="https://www.kaggle.com/c/noaa-right-whale-recognition" target="_blank" rel="nofollow noopener noreferrer">Right Whale Recognition | Kaggle</a> | #-Link-Snipped-#
Christin Khan, a member of NEFSC, under NOAA, works on aerial surveys that capture snapshots of whales from the top, said that identifying whales using mere photography was tedious and there was a high chance of error. To overcome the problem, the NOAA decided to use the technique of facial recognition.
Mother right whale with her calf
Right whales are identified as an endangered species in their North Atlantic habitat. With less than 500 whales, it is crucial to properly label each one so that the total count can be constantly monitored along with their individual health. According to the press release, it is very difficult for even the most experienced researchers to tell whales apart simply by seeing them. Whale faces bear strikingly similar features, among them right whales are those which have raised tissues, dubbed as callosities on top of their heads.
The underlying principle behind the face recognition software was to uniquely categorize callosities or lice patterns. The challenge for data scientists was to automate the right whale recognition process by using given datasets of aerially photographed whales. After a 4-month period, deepsense.io, a software engineering firm overpowered other competitors and grabbed the $10,000 prize. Polish creator Robert Bogucki explained that his team’s whale algorithm would need more data points to improve its precision as the dataset only had 4500 whale photographs.
Right whale identification
Christin claimed that the software would be incorporated in their searching methods within the next two years. This new advancement in their searching technique would definitely enhance the identification process of endangered whales.
Watch Right whale "Calvin" with her calf -
Source: <a href="https://www.kaggle.com/c/noaa-right-whale-recognition" target="_blank" rel="nofollow noopener noreferrer">Right Whale Recognition | Kaggle</a> | #-Link-Snipped-#
0
