With e-commerce marketplaces saturated with high-end brands, cheaper variations of these products have become commonplace over time. Additionally, many consumers have consistently complained about receiving counterfeit products from their online purchases, leading to various policy actions.
Fake products are becoming ubiquitous across categories – from electronics and furniture to apparels and cosmetics. But is there a way to detect counterfeit products before they reach a customer, or even listed on a website?
Sign up for your weekly dose of what's up in emerging technology.
A slew of initiatives like Amazon-led AntiCounterfeiting Coalition (IACC) and Alibaba Big Data Anti-Counterfeiting Alliance have come up in recent years to combat this problem. What is more, numerous startups have been using AI-powered image and text analytics to create solutions that enable stakeholders to discover, thereby curbing this menace.
Fighting Counterfeit Products With AI-Powered Image Analytics
Companies like Seattle-based DataWeave and Pune-based Neurotags have come up with unique solutions that make it easier to authenticate products to ensure only original items reach the market.
DataWeave, for instance, uses deep learning (DL) to detect and eliminate fake products from e-commerce websites. Launched in April 2018, its AI-powered counterfeit products detection solution is trained on millions of catalogue images of products using NVIDIA GeForce GTX 1080 and Caffe deep learning frameworks.
In a blog post, Anshul Garg, a former AI-specialist at the company, illustrated how it uses technology to identify product attributes in the fashion category. Named ‘Fashion Tagger’, it is programmed to automatically assign labels to products at great granularity. In the image of a fashion blogger given below, the tool generated the following output:
According to the company, the algorithm enables easy detection of counterfeit products in catalogue images using its proprietary solution. Garg explains this in three simple steps:
- First train machines to take in its surrounding images using neural-network-based object detection and segmentation.
- These images are then converted to 0s and 1s, and fed into a neural network trained on millions of images acquired from various sources.
- To augment this further, text information – if any – is also processed to enhance the accuracy of the output. This may include non-visual cues like the type of fabric in product descriptions, etc.
This system reportedly has an accuracy of over 95% in detecting counterfeit products. Little wonder then that DataWeave was one of the ‘Best 50 Innovative Applications of AI’ by Nasscom for this solution months after it was launched.
“Our customers across the globe wield our technology to compare their products against those of their competitors,” says Garg. “Our tools provide great granular product-attribute-wise comparisons, enabling curbing of counterfeit products in the market,” he adds.
And while DataWeave’s solution has been serving its clients well over the last two years, NeuroTags has taken a different approach that allows brands to track their product as they travel through the supply chain, until it reaches the consumer.
According to a report, the company provides tags that can be scanned using a smartphone’s native capabilities to get information about a product, which can further be used to ascertain its authenticity. There are two kinds of tags – open and protected. While anyone can scan the open tag which is visible on the product, protected tags can be accessed after a product has been purchased. The tags are connected by an algorithm and AI on the cloud – if anyone tries to replicate them, the copied product gets invalidated, thus, ensuring that counterfeit products are eliminated.
How does it work? The company provides products in its network a unique ID in the form of QR code or RFID tag. Each of these is continuously monitored by an AI-backed cloud, whereby its journey from warehouse to customers is recorded in a dashboard for analysis.
Built by an experienced team from Google, Amazon, PayPal, Visa, and Yahoo, this AI monitored patent-pending anti-counterfeit technology enables consumers to scan products to determine its authenticity before a purchase is made.