Qti.ai enforces intellectual property rights using Oracle AI

Qti.ai protects brands across e-commerce platforms by helping them enforce their intellectual property rights with the following processes: 

• Detecting infringements on clients’ images
• Collecting robust evidentiary data about the scammers
• Removing unauthorized product images
• Pursuing legal action against the infringers and counterfeiters

Qti.ai currently focuses on independent artists and designers, a market largely ignored by most agencies. Qti.ai can provide brand protection services at a reasonable monthly fee, thanks to its proprietary automated systems. The model scales seamlessly, and businesses of all sizes find Qti.ai accessible and affordable.

Qti.ai’s goals

Before their adoption of Oracle Cloud Infrastructure (OCI) technology, Qti.ai individually labeled each image that came through their system, which took several minutes for each infringing listing for the level of accuracy that they needed. Although they sustained this model for a few clients during the proof-of-concept phase, it proved unsustainable as their growth accelerated. 

The amount of time between discovery and removal of an infringing listing is the most critical metric by which some of their clients judge the platform’s performance. The faster they process a listing, the less time consumers are exposed to the scam and less likely to purchase the counterfeit, which ultimately (and rightfully) redirects revenue to legitimate clients of Qti.ai. Before OCI, the documentation and labeling process was unnecessarily onerous.

Legally, a human being needs to confirm product image infringements as a condition of submitting takedown requests, according to the Digital Millennium Copyright Act (DCMA) passed by the US Congress in 1998. However, room for improvement before that stage existed in the Qti.ai process. The company filled that gap by engaging machine learning (ML) tools to scan e-commerce platforms, detect suspicious retail listings, and pre-sort the images with the highest probability of infringing on clients’ brands. Qti.ai adopted several OCI products to improve their process.

Key OCI products used

• OCI Data Labeling: For building labeled datasets to more accurately train artificial intelligence (AI) and ML models. With OCI Data Labeling, developers and data scientists assemble data, create, and browse datasets and apply labels to data records through user interfaces and public APIs.
• OCI Vision: Perform deep learning-based image analysis at scale. With prebuilt models available out of the box, developers can easily build image recognition and text recognition into their applications without ML expertise.
• OCI Object Storage: Enables customers to securely store any type of data in its native format. With built-in redundancy, Object Storage is ideal for building modern applications that require scale and flexibility because it can consolidate multiple data sources for analytics, backup, or archive purposes.

Technical implementation on OCI

Figure 1: A graphic depicting the Qti.ai brand protection process.

Qti.ai uses OCI Vision’s custom models when detecting infringements in cases where scammers modify the original product image by cropping, rotating, inverting colors, or removing / replacing watermarks. To start, the team within Qti.ai generates training sets of infringed images using Data Labeling. When they meet the threshold of accurately labelled data, the team trains and implements a custom Vision model. Then those models were implemented for each brand, so that new images are identified and sorted to gauge the probability of infringement. The prioritization of these images dramatically boosts efficiencies of the review team.

Results

With OCI AI services, Qti.ai was able to label and review large volumes of images within hours, about 5.5 times faster than before migration. The company then reallocated their resources into spending more time interfacing with clients for pursuing legal action against the infringers and counterfeiters, helping their clients’ brands, reputations, and revenue streams. After implementing the Vision service models, even minor changes were identified, labeled, and sorted by probability of infringement, even before reaching the human reviewer.  

Closing thoughts

This post showcases how Qti.ai adopted OCI AI services and automated much of their core functions. Their teams can now focus on servicing and expanding their offerings to their customers.

For more information on Oracle Cloud Infrastructure AI services, see the following resources:

• OCI Data Labeling
• OCI Vision
• LiveLabs