A few thoughts about how AI Pose Detection and Object Detection might be used in the future.
Artificial Intelligence permeates the entire MATRIX blockchain ecosystem. Besides AI-integration for the purpose of intelligent contract security and auto-coding intelligent contracts , the launch of the MATRIX testnet has introduced the MATRIX AI Server. Over the past several days, a number of MATRIX community members have had the opportunity to test the MATRIX AI Server’s AI Pose Detection and Object Detection services. Their experiences have raised a few recurring questions. One is of particular interest.
“Why” is an important question, to be sure. There are a few elements to unpack.
First — Why do these services exist?
The AI Pose Detection and Object Detection have been chosen as the first to roll out on the MATRIX blockchain as they are relatively lightweight and easy to conceptualize. Distributing them on the blockchain is a convenient way to allow the entire community to participate in the training and further refining of these AI Services. This kind of distributed training is more effective than if all training were to be done internally.
Second — Are they services simply cool demonstrations?
Processing these kinds of AI services via blockchain transactions will allow to team to fine-tune the Matrix AI Server. Creating something akin to a live real-world environment will reveal where the AI Server’s shortcomings lie and determine how best to integrate the AI Server in the MATRIX blockchain.
Third — Will these services enable future use-cases?
This may be the first opportunity some MATRIX community members have had to interact with AI services first hand. The hope is that directly interacting with these services will inspire the community to envision how the underlying concepts can be applied in real-world scenarios. To get the ball rolling, a few MATRIX team members were prompted to share their thoughts on how they envision the MATRIX AI Pose Detection and Object Detection services being applied in the future.
However, before sharing their thoughts, a disclaimer is required: The following examples are just the reflections of a few individual MATRIX team members. These examples should not be interpreted as indications that the MATRIX team is currently developing these kinds of applications. Further development of the underlying AI and related technologies, as well as DApp developer support, will be needed if these types of applications are to come to fruition.
Use-Case 1: The Gift of Sight
While large swaths of manufacturing production lines are already automated, human intervention is still often required during key moments due to people’s superior pattern recognition skills. MATRIX AI Object Detection will eventually be able to give machines the gift of sight and continue the march towards automation.
Specifically, MATRIX AI Object Detection could be used to improve defect detection systems in product lines. For example, consider standardized products or product components. AI Object Detection could be trained to detect minuscule differences or variances in the specifications of custom objects and automatically remove defective elements from a production line. MATRIX AI Object Detection could also be useful in the management of completed physical goods. An inventory management system could leverage AI Object Detection to keep count of the number of items in stock. Once an item is removed, the system could automatically detect the change in the physical space and update the inventory accordingly. In this way, the AI Object Detection can remove human error.
This principle could also be scaled down to serve individuals. For instance, it could enable real-time accounting of every item in a person’s pantry. Once a person is running low on watermelon, an application could be made to automatically notify the user or even order more.
Use-Case 2: Push it!
MATRIX AI Pose Detection could be used in a variety of workout applications. It would be relatively straight-forward to develop an application wherein users purchase pre-set workout videos. As users complete their workouts, the application could use the AI Pose Detection to grade and evaluate a user’s performance by tracking and measuring their form, speed or other metrics. Perhaps users can even be rewarded for how well they perform via tokens to unlock the next level of workout programs.
More passive health applications are also easy to imagine. A home office sensor that can trigger a warning when your posture deteriorates or when you’ve been sitting too long could be developed. Of course, similar solutions currently exist in other forms, but AI Pose Detection will enable further optimization. People won’t be able to shake their step counters to get over 10,000 anymore.
Use-Case 3: Health Literacy
Imagine a doctor uploading an image or a scan of their patient’s lungs. This image could serve as a kind of baseline. Then, on a regular basis, a doctor could upload newer scans. The MATRIX AI Object Detection could enable applications to notice changes as they appear. For instance, what if a growth suddenly appears? Even a small change could be automatically detected. With enough training, the MATRIX AI Object Detection could also be able to differentiate between cancerous or benign tumors. The impact of certain experimental treatments could also be tracked and measured.
Depending on the application design, there are vast implications for researchers and practicing physicians. Imagine a shared database of all these scans — each automatically identified and tagged as healthy, cancerous or otherwise by the MATRIX AI Object Detection. Other information about the user’s age, gender, weight and other metrics could also be recorded. This database could be shared with researchers around the world. Additionally, this database could aid physicians in their daily treatment of patients.
The MATRIX AI Object Detection, AI Pose Detection and other unannounced features don’t work in a vacuum, they are tools that can be combined and used to elevate real world applications. However, these kinds of AI processes can also be incredibly power-intensive. This is one of the many contributions of blockchain technology. The shared computing power enabled by blockchain technology has a democratizing effect on the ability of individuals to interact with AI technology. Additionally, as the MATRIX AI Network grows, this vast computing power will enable AI applications beyond the scope of current limitations. The relationship between blockchain and AI is certainly a topic worthy of discussion. More will be shared about these sibling-technologies’ relationship in the coming weeks and months.
Imagining how these applications might shape the world is an inspiring endeavor. While clever people in the MATRIX community and around the world are sure to come up with even better ideas, the above examples hopefully assuage concerns that the AI Object Detection and Pose Detection are simply fun distractions. Rather, they are the early building blocks of useful, real-world applications.