Business
Object Recognition for Price Matching

Source: Finance Derivative
Author: Aleksandras Šulženko, Product Owner at Oxylabs
Pricing intelligence rests on two foundational principles: product and price matching. Extracting the latter data is mostly relatively easy. It usually is easily searchable through the HTML file, making web scraping perfect at picking up price data at scale.
Product matching is where the process becomes complicated. At first glance, it may not seem difficult at all. Simply match the titles across websites, and you’re done. Unfortunately, such an approach would work for a few percentage points of products out of the entire ecommerce industry.
There’s no industry standard on how to create product titles. Additionally, on large user-generated marketplaces, SEO and other marketing considerations might come into play, making it even more challenging to find a perfect match.
Current solutions to product matching
As dynamic pricing is such a popular and important part of ecommerce, several solutions have emerged to tackle the problem. None of them, however, provide a foolproof detection method and are usually used in conjunction.
UPC, EAN, and GTIN comparisons are the most effective by far. They would be almost completely foolproof if not for the fact that few retailers ever publish them. Matching them is preferred to most other methods, but expectations are often shattered due to the scarce availability of such data.
Scraping product specifications such as dimensions, models, production dates, etc. These values are usually static across many retailers as they come from manufacturers and cannot be changed. Slight issues arise as the structure in the way the specifications are displayed isn’t equal across retailers. Additionally, some of them might not list all of the same details.
Finally, there’s the possibility of producing logic trees. Descriptive features (e.g., phone) extracted from categories can be continually matched by other important aspects to create a logic tree (e.g., phone -> iPhone -> iPhone 12 -> iPhone 12 256GB, etc.).
Logic trees greatly reduce the likelihood of false positives but have the drawback of providing fairly few true positives. So, in the end, all methods are usually combined to maximize the probability of matching products.
Object recognition
An understudied area of ecommerce analytics is object recognition. AI of this sort made rounds online about a decade ago as it could separate cats from dogs, the internet’s favorite image source. Since then, significant strides have been made in the development of AI object recognition.
It could have its fair uses in product matching for ecommerce. Most retailers are heavily invested in high-quality images (or, in some cases, required to provide them) with clearly stated branding. A fair part of boxed products will have the product’s name listed on them with some potential for a description.
Machine learning models can derive fairly accurate descriptions of objects without any additional descriptors. Fine-tuned ones would be able separate objects out of categories, and ones dedicated to specific categories would be able to differentiate between objects within them.
Since most products, however, have descriptors added to the packaging or images, such as the aforementioned titles or other words, those can be also extracted. Marketing practices state that essential, differentiating information should be displayed most prominently, allowing a machine learning model to bypass other data collection methods.
Although there are some caveats, most prominently, not all products can be differentiated purely by image. For example, iPhone versions could be detected, but it’s impossible to extract in-built storage capabilities (i.e., 256 GB vs 512 GB) out of the image. Therefore, in some cases, other sources will have to be used.
Additionally, some products may be extremely similar between themselves (such as some of the IKEA range), which, even with a well-trained and adapted machine learning model, may be hard to detect outright.
Increasing accuracy
There are some inherent benefits in ecommerce product object recognition. Retailers have the incentive to create crisp quality images that clearly showcase specific products as it improves conversion rates.
In many cases, recognition deals with images of highly variable quality, angles, and primary object visibility. In ecommerce, many of these issues will be less prevalent due to the reasons outlined above.
Yet, there are still plenty of reasons to improve accuracy, as every percentage point will have a compounding effect in the long run. One of the options is to collect more data, which always works, however, it’s not the only way.
Data augmentation, the practice of tinkering with existing data to create new points, is perfectly suited for object recognition. Unlike text-based and numerical data, images can have nearly an infinite number of small changes while retaining the original intention.
Common examples of data augmentation include object occlusion, using photometric or geometric distortion (i.e., changing brightness, cropping, etc.), and superimposing two or more images on top of each other.
Object occlusion has shown promising results in making models more accurate at making predictions. The running theory is that by occluding certain parts of an object, the model is forced to focus on other parts to make a prediction, eliminating some possible skew.
Outside of object recognition practices, the model can be integrated with existing product matching systems. Each prediction can be matched in with an existing product in the database to see whether the specifications and other details also align.
For example, a prediction about a specific type of iPhone might turn out to be erroneous because the dimensions of that model don’t add up. In other words, there are some “hard facts” about products that never change. They can be used to hedge predictions to ensure that the system comes up with a higher level of accuracy.
So, it may seem like simply another method that could be viable at detecting and matching products. Yet, there is something important in relation to machine learning and web scraping.
Web scraping builds models
One of the hardest parts, if not truly the most complicated, is getting all the data that’s needed for a machine learning model. Typically, you’d have to scrape thousands of pages, label data, and keep feeding it into the algorithm.
But wherever pricing intelligence is already in place, the data is readily available. All the other methods of product matching rely on procuring the data that could be easily used to build a machine learning model.
As such, the resource costs associated with creating one are minimized. There still would have to be some sort of labeling involved, however, even that could be automated. After all, the products are already matched, so the desired output is known.
Since web scraping almost always downloads the entire HTML and parses it through to deliver the necessary data, downloading images to feed into the algorithm isn’t much of a change to the regular course of action. One word of caution, however, is that image delivery would greatly increase traffic costs for proxies, which can affect overall operation costs.
Therefore, the supplementary model is almost already available. Most of the hard work required to create one is done by the requirements of pricing intelligence. As such, gathering a dataset for implementing object recognition for price and product matching is much simpler than it may seem at first glance.
Conclusion
Product matching is likely one of the most complicated tasks allotted to ecommerce analytics. While rarely used, object recognition is one way to increase the likelihood of true positive detection.
One question remains – how much should one trust the model’s output? Unfortunately, I believe there’s no easy decision as accuracy is dependent on so many factors that giving a blanket answer is meaningless.
You may like
Business
What can the West learn from the Arabian Gulf’s payments revolution?

Hassan Zebdeh, Financial Crime Advisor at Eastnets
A decade ago, paying for coffee at a small café in Riyadh meant fumbling with cash – or, at best, handing over a plastic card. Today, locals casually wave smartphones over terminals, instantly settling the bill, splitting it among friends, and even transferring money abroad before their drink cools.
This seemingly trivial scene illustrates a profound truth: while the West debates incremental upgrades to ageing payment systems, the Arabian Gulf has leapfrogged straight into the future. As of late 2024, Saudi Arabia achieved a remarkable 98% adoption rate for contactless payments in face-to-face transactions, a significant leap from just 4% in 2017.
Align financial transformation with a bold national vision
One milestone that exemplifies the Gulf’s approach is Saudi Arabia’s launch of its first Swift Service Bureau. While not the first SSB worldwide, its presence in the Kingdom underscores a broader theme: rather than rely on piecemeal upgrades to older infrastructure, Saudi Arabia chose a proven yet modern route, aligned to Vision 2030, to unify international payment standards, enhance security, and reduce operational overhead.
And it matters, because in a region heavily reliant on expatriate workers whose steady stream of remittances powers whole economies. The stakes for frictionless cross-border transactions are unusually high. Rather than tinkering around the edges of an ageing system, Saudi Arabia opted for a bold and coherent solution, deliberately aligning national pride and purpose with practical financial innovation. It’s a reminder that infrastructure, at its best, doesn’t merely enable transactions; it reshapes how people imagine the future.
Make regulation a launchpad, not a bottleneck
Regulation often carries the reputation of an overprotective parent – necessary, perhaps, but tiresome, cautious to a fault, and prone to slowing progress rather than enabling it. It’s the bureaucratic equivalent of wrapping every new idea in bubble wrap and paperwork. Yet Bahrain has managed something rare: flipping the narrative entirely. Instead of acting solely as gatekeepers, Bahraini regulators decided to become collaborators. Their fintech sandbox isn’t merely a regulatory innovation; it’s psychological brilliance, transforming a potentially adversarial relationship into a partnership
Within this curated environment, fintech firms have launched practical experiments with striking results. Take Tarabut Gateway, which pioneered open banking APIs, reshaping how banks and customers interact. Rain, a cryptocurrency exchange, tested compliance frameworks safely, quickly becoming one of the Gulf’s trusted crypto players. Elsewhere, startups trialled AI-driven identity verification and seamless cross-border payments, all under the watchful yet adaptive guidance of Bahraini regulators. Successes were rapidly scaled; failures offered immediate lessons, free from damaging legal fallout. Bahrain proves regulation, thoughtfully applied, can genuinely empower innovation rather than restrict it.
Prioritise cross-border interoperability and unified standards
Cross-border payments have long been a maddening puzzle – expensive, sluggish, and unpredictably complicated. Most Western banks seem resigned to this reality, treating the spaghetti-like mess of correspondent banking relationships as a necessary evil. Yet Gulf states looked at this same complexity and saw not just inconvenience, but opportunity. Instead of battling against the tide, they cleverly redirected it, embracing standards like ISO 20022, which neatly streamline data exchange and slash friction from global transactions.
Examples abound: Saudi Arabia’s adoption of ISO 20022 through its Swift Service Bureau will notably accelerated cross-border transactions and improve transparency. The UAE and Saudi Arabia also jointly piloted Project Aber, a digital currency initiative that significantly reduced settlement times for interbank payments. Similarly, Bahrain’s collaboration with fintechs has simplified previously burdensome remittance processes, reducing both cost and complexity.
Target digital ecosystems for financial inclusion
One of the most intriguing elements of the Gulf’s payments transformation is the speed and enthusiasm with which consumers embraced new technologies. In Bahrain, mobile wallet payments surged by 196% in 2021, contributing to a nearly 50% year-over-year increase in digital payment volumes. Similarly, Saudi Arabia experienced a near tripling of mobile payment volumes in the same year, with mobile transactions accounting for 35% of all payments.
The West, by contrast, still struggles with financial inclusion. In the U.S., millions remain unbanked or underbanked, held back by distrust, geographic isolation, and high fees. Digital solutions exist, but widespread adoption has lagged, partly because major institutions view inclusion as a long-term aspiration rather than an immediate priority. The Gulf shows that when digital tools are made integral to daily life, rather than optional extras, the barriers to financial inclusion quickly dissolve.
The road ahead
As the Gulf region continues to refine its payment systems experimenting with digital currencies, advanced data protection laws, and AI-driven compliance the ripple effects will be felt far beyond the GCC. Western players can treat these developments as an external threat or as a chance to rejuvenate their own approaches.
Ultimately, if you want a glimpse of where financial services may be headed towards integrated platforms, real-time international transactions, and widespread digital inclusion – the Gulf experience is a prime example of what’s possible. The question is whether other markets will step up, follow suit, and even surpass these achievements. With global financial landscapes evolving at record speed, hesitation carries its own risks. The Arabian Gulf has shown that bold bets can pay off; perhaps that’s the most enduring lesson for the West.
Business
Unlocking business growth with efficient finance operations

Rob Israch, President at Tipalti
The UK economy has faced a turbulent couple of years, meaning now more than ever, businesses need to stay agile. With Reeves’s national insurance hikes now fully in play and global trade tensions casting a shadow over the landscape, the coming months will present a crucial opportunity for businesses to decide how to best move forward.
That said, it’s not all doom and gloom. The latest official figures show that the UK’s economy unexpectedly grew at a rate of 0.5% in February – a welcome sign of resilience. But turning this momentum into sustainable growth will hinge on effective financial management – essential for long term success.
Although many are currently prioritising stability, sustainable growth is still within reach with the right approach. By making use of data and insights from the finance team, companies can pinpoint efficient paths to expansion. However, this relies on having real-time information at their fingertips to support agile, well-timed decisions.
While achieving growth may be tough to come by this year, businesses can stay on track by adopting a few essential strategies.
Improving efficiency by eliminating finance bottlenecks
Growth is the ultimate goal for any business, but it must be managed carefully to ensure long-term sustainability. Uncertain times present an opportunity to eliminate inefficiencies and build a strong foundation for future success.
A significant bottleneck for many businesses is the finance function’s reliance on manual processes for invoice processing, reporting and reconciliation. These tasks are not only time-consuming but also introduce errors, delays and inefficiencies. As a result, finance teams become stretched thin. Our recent survey found that, on average, over half (51%) of accounts payable time is spent on manual tasks – severely limiting finance leaders’ ability to drive strategic growth.
Repetitive tasks such as data entry, reconciliation, and approvals require considerable time and effort, slowing down decision-making and increasing the risk of inaccuracies. Given the critical role that finance plays in guiding business strategy, these inefficiencies and errors create significant roadblocks to growth.
The pressure on finance leaders is therefore immense and while 71% of UK business leaders believe CFOs should take a central role in corporate growth initiatives, they are simply lost in a sea of manual processes and number crunching. In fact, 82% of finance leaders admit that excessive manual finance processes are hindering their organisation’s growth plans for the year ahead. To remedy this, businesses must embrace automation.
Achieving sustainable growth with automation
By replacing manual spreadsheets with automated solutions, finance teams can eliminate administrative burdens and focus on strategic initiatives. Automation simplifies critical finance tasks like bank feeds, coding bookkeeping transactions and invoice matching. Beyond this, it can also help alleviate the strain of more complex and time-intensive responsibilities, including tax filings, invoices and payroll.
The benefits of automation extend far beyond time saving, to accuracy, improving business visibility and enabling real-time financial insights. With fewer errors and faster-data processing, finance leaders can shift their focus to high-value tasks like driving strategy, identifying risks and opportunities and determining the optimal timing for growth investments.
Attracting investors with operational efficiency
Once businesses have minimised time spent on administrative tasks, they can focus on the bigger picture: growth and securing investment. With access to cheap capital becoming increasingly difficult, businesses must position themselves wisely to attract funding.
Investors favour lean, efficient companies, so demonstrating that a business can achieve more with fewer resources signals a commitment to financial prudence and sustainability. By embracing automation, companies can showcase their ability to manage operations efficiently, instilling confidence that any new investment will be spent and used wisely.
Economic uncertainty provides an opportunity to reassess business foundations and create more agile operations. Refining workflows and eliminating bottlenecks not only improves performance but also strengthens investor confidence by demonstrating a long-term commitment to financial health.
Additionally, strong financial reporting and effective cash flow management are crucial to standing out to investors. Clear, real-time insights into financial health demonstrate resilience and highlight a business’ resilience and readiness for growth.
The growth journey ahead
Though the landscape remains tough for UK businesses, sustainable growth is still achievable with a clear and focused strategy. By empowering finance leaders to step into more strategic and high-level decision making roles, organisations can stay resilient and agile amid ongoing economic headwinds.
UK businesses have fought to stay afloat, so now is the time to rebuild strength. By embracing more strategic financial management to build resilience, they can set the stage for long-term, sustainable growth, whatever the economic climate brings.
Business
The Consortium Conundrum: Debunking Modern Fraud Prevention Myths

By Husnain Bajwa, SVP of Product, Risk Solutions, SEON
As digital threats escalate, businesses are desperately seeking comprehensive solutions to counteract the growing complexity and sophistication of evolving fraud vectors. The latest industry trend – consortium data sharing – promises a revolutionary approach to fraud prevention, where organisations combine their data to strengthen fraud defences.
It’s understandable how the consortium data model presents an appealing narrative of collective intelligence: by pooling fraud insights across multiple organisations, businesses hope to create an omniscient network capable of instantaneously detecting and preventing fraudulent activities.
And this approach seems intuitive – more data should translate to better protection. However, the reality of data sharing is far more complex and fundamentally flawed. Overlooked hurdles reveal significant structural limitations that undermine the effectiveness of consortium strategies, preventing this approach from fulfilling its potential to safeguard against fraud. Here are several key misconceptions about how consortium approaches fail to deliver promised benefits.
Fallacy of Scale Without Quality
One of the most persistent myths in fraud prevention mirrors the trope of enhancing a low-resolution image to reveal more explicit details. There’s a pervasive belief that massive volumes of consortium data can reveal insights not present in any of the original signals. However, this represents a fundamental misunderstanding of information theory and data analysis.
To protect participant privacy, consortium approaches strip away critical information elements relevant to fraud detection. This includes precise identifiers, nuanced temporal sequences and essential contextual metadata. Through the loss of granular signal fidelity required to anonymise information to make data sharing viable, said processes skew data while eroding its quality and reliability. The result is a sanitised dataset that bears little resemblance to the rich, complex information needed for effective fraud prevention. Further, embedded reporting biases from different entities can likewise exacerbate quality issues. Knowing where data comes from is imperative, and consortium data frequently lacks freshness and provenance.
Competitive Distortion is a Problem
Competitive dynamics can impact the efficacy of shared data strategies. Businesses today operate in competitive environments marked by inherent conflicts, where companies have strategic reasons to restrict their information sharing. The selective reporting of fraud cases, intentional delays in sharing emerging fraud patterns and strategic obfuscation of crucial insights can lead to a “tragedy of the commons” situation, where individual organisational interests systematically degrade the potential of consortium information sharing for the collective benefit.
Moreover, when direct competitors share data, organisations often limit their contributions to non-sensitive fraud cases or withhold high-value signals that reduce the effectiveness of the consortium dynamics.
Anonymisation’s Hidden Costs
Consortiums are compelled to aggressively anonymise data to sidestep the legal and ethical concerns of operating akin to de facto credit reporting agencies. This anonymisation process encompasses removing precise identifiers, truncating temporal sequences, coarsening behavioural patterns, eliminating cross-entity relationships and reducing contextual signals. Such extensive modifications limit the data’s utility for fraud detection by obscuring the details necessary for identifying and analysing nuanced fraudulent activities.
These anonymisation efforts, needed to preserve privacy, also mean that vital contextual information is lost, significantly hampering the ability to detect fraud trends over time and diluting the effectiveness of such data. This overall reduction in data utility illustrates the profound trade-offs required to balance privacy concerns with effective fraud detection.
The Problem of Lost Provenance
In the critical frameworks of DIKA (Data, Information, Knowledge, Action) and OODA (Observe, Orient, Decide, Act), data provenance is essential for validating information quality, understanding contextual relevance, assessing temporal applicability, determining confidence levels and guiding action selection. However, once data provenance is lost through consortium sharing, it is irrecoverable, leading to a permanent degradation in decision quality.
This loss of provenance becomes even more critical at the moment of decision-making. Without the ability to verify the freshness of data, assess the reliability of its sources or understand the context in which it was collected, decision-makers are left with limited visibility into preprocessing steps and a reduced confidence in their signal interpretation. These constraints hinder the effectiveness of fraud detection efforts, as the underlying data lacks the necessary clarity for precise and timely decision-making.
The Realities of Fraud Detection Techniques
Modern fraud prevention hinges on well-established analytical techniques such as rule-based pattern matching, supervised classification, anomaly detection, network analysis and temporal sequence modelling. These methods underscore a critical principle in fraud detection: the signal quality far outweighs the data volume. High-quality, context-rich data enhances the effectiveness of these techniques, enabling more accurate and dynamic responses to potential fraud.
Despite the rapid advancements in machine learning (ML) and data science, the fundamental constraints of fraud detection remain unchanged. The effectiveness of advanced ML models is still heavily dependent on the quality of data, the intricacy of feature engineering, the interpretability of models and adherence to regulatory compliance and operational constraints. No degree of algorithmic sophistication can compensate for fundamental data limitations.
As a result, the core of effective fraud detection continues to rely more on the precision and context of data rather than sheer quantity. This reality shapes the strategic focus of fraud prevention efforts, prioritising data integrity and actionable insights over expansive but less actionable data sets.
Evolving Into Trust & Safety: The Imperative for High-Quality Data
As the scope of fraud prevention broadens into the more encompassing field of trust and safety, the requirements for effective management become more complex. New demands, such as end-to-end activity tracking, cross-domain risk assessment, behavioural pattern analysis, intent determination and impact evaluation, all rely heavily on the quality and provenance of data.
In trust and safety operations, maintaining clear audit trails, ensuring source verification, preserving data context, assessing actions’ impact, and justifying decisions become paramount.
However, the nature of consortium data, which is anonymised and decontextualised to protect privacy and meet regulatory standards, cannot fundamentally support clear audit trails, ensure source verification, preserve data context, and readily assess the impact of actions to justify decisions. These limitations showcase the critical need for organisations to develop their own rich, contextually detailed datasets that retain provenance and can be directly applied to operational needs to ensure that trust and safety measures are comprehensive, effectively targeted, and relevant.
Rethinking Data Strategies
While consortium data sharing offers a compelling vision, its execution is fraught with challenges that diminish its practical utility. Fundamental limitations such as data quality concerns, competitive dynamics, privacy requirements and the critical need for provenance preservation undermine the effectiveness of such collaborative efforts. Instead of relying on massive, shared datasets of uncertain quality, organisations should pivot toward cultivating their own high-quality internal datasets.
The future of effective fraud prevention lies not in the quantity of shared data but in the quality of proprietary, context-rich data with clear provenance and direct operational relevance. By building and maintaining high-quality datasets, organisations can create a more resilient and effective fraud prevention framework tailored to their specific operational needs and challenges.

What can the West learn from the Arabian Gulf’s payments revolution?

Unlocking business growth with efficient finance operations

The Consortium Conundrum: Debunking Modern Fraud Prevention Myths

Stealthy Malware: How Does it Work and How Should Enterprises Mitigate It?

Future-proofing the workforce for AI innovations with continuous learning
