A Singapore-based firm has employed computer vision technology in a mobile service designed to improve efficiency in the retail sector.
Called Trax, the service reduces by 60 per cent the time an employee needs to spend on audits to check inventory, shelf display and product promotions.
A mobile phone or tablet is used to photograph the products on the shelves. The Trax algorithm based on computer vision technology recognises the products and the way they are displayed.
It captures the data and gives an analysis within 10 minutes.
Trax co-founder and chief executive Joel Bar-El said: "Generally, the algorithm can determine the availability, assortment, shelf space, pricing, promotions, shelf location and special arrangements like arranging a product display from left to right according to price.
"With the manual systems, the retail executive has to physically measure the space devoted to the product, and count inventory."
Only four to five factors can be captured manually, compared to the algorithm's ability to automatically measure 100 factors.
Information collected on global brands from different countries is also useful. One country could have benefited from a merchandising display method, so another could try the same method.
Trax allows employees to cover more supermarkets and outlets in a day. Previously they could cover only two places a day as it would take several hours to manually audit one store.
Mr Bar-El said: "Market intelligence of this type is critical for fast-moving consumer goods because they pay for space in supermarkets and stores.
"In the United States alone, these companies pay about US$18 billion (S$25 billion) for shelf space. So the analytics we provide helps them sell more products."
Founded five years ago, Trax now has more than 40 customers in the fast-moving consumer goods industry, including beverage giant Coca- Cola and brewer Anheuser-Busch InBev as global customers. Its service is available in 32 markets.
"We have competitors but they provide analysis in 24 hours, whereas we provide results immediately," said Mr Bar-El, an Israeli who is a Singapore permanent resident.
The main challenge in the 18 months spent developing Trax concerned fine-grain recognition.
"It is easy to recognise a person. There's a head, two eyes, a nose and a mouth. Products like Coca-Cola come in different sizes as well as in different packaging like bottles and cans. So we needed to recognise products that are similar or identical, yet be able to differentiate them," said Mr Bar-El.
In the beginning, the company piloted its technology with customers. This allowed the algorithm to "learn" about different products and the more it "learns" the better it gets in recognising same products in different shapes and sizes.
Trax has its headquarters in Singapore, and a development centre in Tel Aviv.
Mr Bar-El added: "We hope to raise about US$40 million for product development as well as beef up our sales and marketing team and open new offices."
