We’re always excited about the opportunity to work with smart and innovative projects that have the potential to bring a completely new dimension to certain activities.
At Pecode, we look forward to building long-term partnerships based on trust and common aspiration to get to new heights.
In this case study, we present the outcome of our cooperation with an AI development company for the work on their AI SDK that tracks human emotions.
Brief summary
- Our main task was to develop a library interface and documentation using MediaPipe. We also had to perform deployment and testing of the finished solution.
- Our team helped the client to develop mobile apps for Android and iOS.
- We successfully finished the confirmed scope of work and continued our collaboration for the solution’s ongoing maintenance.
About the project
Our client is developing an emotional well-being AI to help users improve mental health using high-end technology. Their emotional AI SDK measures the intensity and duration of multiple human micro-expressions using a built-in video camera. The solution provides seamless mobile and web experiences, enables real-time analysis, and more.
The client’s request
The primary request from our client was to develop a comprehensive library interface along with detailed documentation. Our team also covered the development of mobile applications for Android and iOS.
We were responsible for the deployment and testing processes as well as post-launch bug fixes and new features implementation.
The challenge that we faced
The biggest challenge for us was to ensure that the library would work seamlessly for third-party developers. To achieve this, we implemented a specialized testing approach that encompassed the following critical steps:
- Integration of the library into a third-party project
- Documentation validation
The correctness and clarity of the documentation were paramount to ensure a smooth user experience for third-party developers. We conducted thorough reviews and walkthroughs of the documentation to identify and promptly address any ambiguities or gaps.
Solutions we provided
We developed and integrated several advanced features within the library, all aimed at enhancing the functionality and accuracy of the solution’s performance. Specifically, we focused on the following key areas:
- Integration of face blend shapes that allow for capturing and processing various facial expressions to create a detailed and dynamic model of the user’s face.
- Implementation of a head transformational matrix to track and analyze head movements in three-dimensional space.
- Improvement of computation methods through the development of advanced algorithms.
All the features we implemented are integral components of the ML algorithm pipeline. The ability to collect data using the MediaPipe framework has significantly improved the accuracy and scope of the library.
The successful integration and deployment in practical applications, such as Temi robots, demonstrate the real-world impact and benefits of our work.
Tech Stack
- Mediapipe: both platforms
- Android: Android SDK, Kotlin, Kotlin Coroutines, Java Concurrency
- iOS: iOS SDK, Swift, MediaPipeTasks, Cocoapods
Result
Our joint work with the AI development company was quite successful. We completed the confirmed scope of work within the defined timeframe and due to the client’s request.
We added new innovative features that contributed to a superior user experience. The work on the project is still in progress – the client collects and analyzes user feedback and requests our team to integrate new features or fix bugs when necessary.
