Computer Vision & Image Analysis as Game Changer for both Businesses and Consumers
Why is Computer Vision & Image Analysis a Game Changer?
More than 3 billion photos are shared every day on social media. Understanding the consumption of images that feature a companies’ products provides insight throughout the business. This aspect alone has the potential to disrupt and innovate across every step in the process – namely imagine, create, produce, use, value, advocate.
Today, digital advertising is finally surpassing traditional, offline advertising. 51% of the USA Ad spend of 130 billion, is targeted to create a plethora of product images. However, these digital, product images deliver value primarily restricted to a business’ marketing activities.
What is Computer Vision & Image Analysis?
Computer Vision is a sub-field of artificial intelligence (AI) that uses machine learning to automatically understand any image (or video) based on the contextual visual elements and patterns. Computers can see shapes and more importantly extract meaning from images.
Does it work? The United States FDA has approved software to identify a vision-impairing disease of the retina by reviewing images of the eye. The images are taken with a special camera without the prerequisite of a medical specialist to also review the images. This means, Computer Vision has already transcended the accuracy of specialized & highly trained human vision.
Image Analysis is a sub-field of Data Analysis that pulls relevant information from an image (or video) for advanced classification and analysis.
Potential impact? Think, Amazon, Spotify, Netflix (non-image), and so on. Data Analysis paved the way for innovation and differentiation in the way books, music and entertainment are produced, distributed, and consumed. New business models, new market leaders and the elimination of traditional players have resulted. Here, the intent is to outline why Image Analysis promises to be just as disruptive, if not more so.
The Strategic Value Opportunity
Tomorrow
Computer Vision & Image Analysis, when applied to understanding the consumption of digital product images will provide strategic value throughout business by being able to answer the following questions:
- Which new product features are gaining the most traction?
- Which are the most favorable feature combinations?
- Which trends are most popular and can they be extended?
- Which versions of images resonate best with consumers?
Predicting the Future
Digital product images (often produced from 3D design data), published early in the ideation phase, allow for feedback on design decisions before the product has been built and is in the hands of the customer. This allows companies to freely test their ideas, their products, and the experiences they provide consumers before actually producing them. This is a radical departure from the existing product design and creation process which today results in 90% of new product launches failing despite expensive prototype production.
Sustainability Contribution
The production of only those goods where a value and need is identified is now possible without wasting critical resources. In addition, there is also a discussion to be had on challenging the fundamentals of the Industrial economy: the increased focus on creating value first rather than just focusing on growth.
Trying to Predict Product Success is Not New
Businesses have traditionally employed costly methods like surveys, focus groups, and small-batch trial production runs to predict product success, albeit with limited accuracy.
So, what’s new?
Computer Vision & Image Analysis can automate the accurate identification of product images ‘in the wild’ (even before the physical product exists) as consumers upload, comment, share and interact with images on social media and the internet.
Identifying where your product is present online and understanding what is being said about it allows for true ‘consumer lead’ innovation. The source of innovation is now the connection between the world of consumption and the world of production.
Quicker, more accurate business decisions result from simplifying access to, and providing an understanding of, image analysis data. Data-driven decisions are a competitive advantage for new product innovation. By understanding product image consumption, any brand can make more accurate decisions, it’s simple.
New market research data points provided by computer Visions & Image Analysis include:
- Measure the number of product image results (total unique images featuring a product)
- Distinguish between various versions of a product (color, variant, versions)
- Find influencers linked to products (YouTubers with >1M followers who feature product)
- Detect product affinities (any product displayed with high-frequency alongside another)
- Identify moments of consumption (where/in what circumstances is a product shown)
- Analyze consumer sentiment displayed with a product (people smiling, laughing, neutral etc)
Computer Vision & Image Analysis-Enabled Innovation Opportunities
Faster Product Cycles
Predict potential product failures and areas for improvement using market research data based on the consumption of product imagery for unreleased products by analyzing it and reacting quickly. A long-term implication includes using prescriptive analytics in product design. Teams can solve problems and shape outcomes proactively, rather than simply predicting and responding to demand in product design.
Agile Product Innovation
Better decisions can be made on a short-cycle experimentation and feedback loop. Designers can respond quickly to apply modifications (on a product or an experience) based on insights and precise tracking of a consumers’ product experience (what worked vs. what didn’t).
Customer Segmentation
Enhanced customer demographic information (who are they) can be supplemented with Psychographic information (what kind of person are they). Things like traits, beliefs, values, attitudes, interests, and lifestyles are often decipherable from images or consumers profiles.
Product Segmentation
Digital product image consumption data provides insights into niche consumer interests and preferences. Smart Factories, an approach to production where the linear assembly line has been replaced by the concept of modular assembly. Product customization is possible with highly flexible (mass) production.
Consumer engagement
This aspect is more important than features and benefits alone. Consumers expect to interact with or even influence brands – not just be sold to. Understanding the customer is where every innovation starts. Consumer feedback fuels ideation and innovation in production. Ultimately, innovative products should also improve the lives of those who purchased them.
Supply Chain Efficiency
Supply chain data can be seamlessly integrated using Image Analysis supported by Computer Vision. The ability to predict demand results in preemptive production and efficient delivery of products to customers.
Consumer Benefits
Computer Vision & Image Analysis
At Zara, store personnel currently procure important, individualized shopper information manually. Customer preferences are captured for every detail: buttons, zippers, color, cut and more. They enter and upload the feedback nightly and along with data from Instagram and other social media, isolate their geographic region's tastes to drive new design releases and ship to stores twice a week. Zara has a deep understanding of their customer and offer designs in tiny batches that eventually run out - but rarely go on sale. When they get it wrong, they adapt faster than their competitors.
Personalized design and production
Instead of uniform production, apparel and consumables can be tailored on-demand. We could eventually move to fully interactive and customized design and supply where digital-only mock-ups of garments are sold online, made in small batches using automated production, and subsequent changes to the next design batch are made based on user feedback.
On-demand customization
By anticipating demand, information about what consumers want, when and how they want it is readily available. Personalized design and production enabled by this evolution will change our expectations. We’re already used to programmatic product recommendations based on our preferences, soon we’ll be recommended products that are fully customized and tailored to us.
Supply chain and production optimization
Consumers benefit indirectly from more flexible, responsive and custom-made product manufacturing. Time savings combined with faster response times, fewer delays, fewer defects and faster deliveries are just a few reasons why. Today, we have on-demand production and increased automation of significant production processes. This will only improve with intelligent automation in areas ranging from supply chain optimization to more predictive scheduling.
Finally, consumers save time
Imagine a reality where exploring shelves, catalogues, and websites becomes irrelevant to finding a desired product. Consumers, attracted to higher quality and more personalized products and services, increase consumption, thereby creating a virtuous cycle of more data touchpoints and more data, better insights and better products.
Successful Implementation Top Tips
Computer Vision & Image Analysis
Early adopters will have the advantage of superior customer insight. The immediate competitive advantages include an improved ability to tap into consumer preferences, customize products to match individual demands to capture an ever bigger slice of the market. Shaping product development around this rich supply of customer data will make it harder and harder for slower moving competitors to keep up and could eventually make their advantage unassailable.
Artificial Intelligence Strategy
Every brand must define how Computer Vision & Image Analysis should be used. For instance, whether to focus on gains in market research or input for product enhancements. The decision will be based on specific circumstances. However, pinpointing a direction will be a game-changer for their business. Equally, identifying subsequent use-cases and business needs to determine which solutions should be deployed require decisive processes. These include, prioritizing the implementation and determining how much and where to invest in Computer Vision & Image Analysis.
Enabling strategic execution
Establishing a virtual, ‘Center of Excellence’ (COE) organizational approach will make Integrating Computer Vision & Image Analysis in the business stress-free. In doing so, the COE can tap into best practices and expert abilities to deploy tactics strategically (enterprise-wide) while staying close to and building relationships with individual business units. The challenge is not in the technology, with many capabilities open-source and readily reusable. The challenge resides in gaining a deep understanding of and possessing the ability to frame business challenges in a way to solve them by the application of Computer Vision & Image Analysis.
Technology make or buy?
Substantial investment in recent years has led to the successful development of several high-quality Computer Vision & Image Analysis software offerings (Amazon Rekognition, Google Vision, and IBM Watson Visual Recognition) that are reasonably priced and available as SaaS. As the market for top-notch data-scientists is exceptionally competitive and likely to remain so, ‘buy’ is the obvious choice for all but the largest technology companies. The AI startup space is also bustling with the most successful organizations choosing one of a few directions. Either they focus on supporting enterprises with consulting services and tooling to integrate AI solutions into their business (Clarifai and DataRobot), or they specialize in niche areas like AI for mobile devices (Pilot.ai), robot cameras (Prophesee), or interpreting satellite data (Orbital Insight).
What’s Your Computer Vision & Image Analysis Use Case?
Do you have a plan to succeed?
How are you estimating and valuing expected benefits (either operational cost savings or additional revenue opportunities) of your use-case? Do you have the right team in place, either internally or externally to deliver? And what steps do you need to take to ensure buy-in and active support from your colleagues and other partners in your ecosystem?
This is the second article in a series from the Dassault Systèmes 3DEXCITE Strategy Department. Get involved in the conversation - like, share, comment!
Previous article: A perspective on the Next Industrial Revolution - The future of work and consumption (by Fabien BARTEL)
