Agora's object
recognition
My role: UX Research & UI Design
​
Team: Maria Fernandez, Elena Malakhina,
Ben Wooten
​
My key tasks: Competitor analysis, survey creation, user journey, user persona, sketching, concept creation.
​
Timeline: Hackathon -24h-
An introduction
About the project
Agora is a video conferencing and online streaming platform that wanted to develop their object recognition
feature within the online community.
​
Our team had in mine the importance of physical and
mental health during this unprecedented times and we
wanted to offer a solution.
The problem
Many members of fitness clubs wanted to
keep healthy during lockdown and they were using live streaming services to follow teacher's classes.
​
Teachers found out that is quite difficult to correct everyone's position during the class without interrupting the flow.
​
Participants also didn't want other members of the class to see them so they turned their cameras off. This means the instructor couldn't check their posture and also made the flow of the class less interactive.
​
The project goal
The goals are:
1. Create a feature were teachers could instruct people without interrupting the whole class.
​
2. Make members feel welcomed and that no one is invaded their own space.
​
Checking competitors
We did a competitor research of these 3 apps:
​
Understanding the problem space
What are the challenges that participants face when they attend online fitness classes?
We conducted twenty surveys and two interview with the target group to understand their needs and pain points when hospitals are overcrowded.
​
The major questions we asked here were:
​
- What has your hospital improved during this pandemic?
- Could an app speed up the current process? How?
- What is the step by step process when a patient arrives?
- What are the biggest challenges you face when you check in patients?
- Is there any frustrations by the way is run?
- If the answer is yes… which one?
- How does Health Insurance affect this process?
​
​
​
These were the main results:
Brainstorming
.jpg)
We wrote our observations on the stickers and put them on the wall. It helped us to quickly made sense
of user research by finding ideas. We have voted for the most important parts of the user experience based on our competitors' analysis and survey results
​
​
​
Exploration
What are the potential features based on the main pain points?
We created a user persona and empathy map to determine potential features. This was also an opportunity for us to further understand their needs and gather insights from what was not explicitly mentioned.
​
Insight and possible features:
​
- Use Agora's recognition feature to check participants posture and offer suggestions without interrupting the flow of the class. It would be just shown of each participants screen.
​
- The participants would have the options to turn their cameras on a private mode (teacher view only) or public mode (all participants view).
​
​
Proposed solutions
Based on the insights from the generative research, surveys, the personas, interviews and user journeys, we further refined the features for the minimum viable product.
​
- Object recognition feature that assists participants correcting their posture.
​
- Option to toggle the camera between
private or public mode
​
​
​
​
Designing high-fidelity prototype
Users may use a private mode.
No one except a teacher can see you.
​
​
​
Timer shows time left in the class.
​
​
​
The next pose window provides a hint for
users. They have time to prepare
​
​
Object recognition technology identifies the
instructor pose and compares it with the user.
Based on that users get personal
recommendations along with the message.
No one will hear feedback about your class
process.
​
​
Class Options menu allows users to
customize class parameters
Feedback
​
​
​
We submitted the project and the feedback was very valuable to the team.
Out of 40 projects that participated we were awarded with a 2nd place from the judges.
​
Problems encountered and solutions angle:
​
​
​
1. Complex onboarding process
Solution: Simplify onboarding with a step-by-step guide and real-time feedback (e.g., "Camera aligned!").
​
​
2. Inaccurate Feedback
Solution: Improve object recognition accuracy and show confidence levels in feedback (e.g., “90% Confidence - Great!”).
​
​
3. No Immediate Positive Feedback
Solution: Provide real-time praise (e.g., “Great job! Your form is perfect”) and gamify the experience with badges or achievements.
​
​
4. Integration with Other Fitness Apps
Solution: Integrate with Fitbit, Apple Health, etc., to track overall progress and improve retention.
​
​
5. Lack of Contextual Guidance.
Solution: Offer exercise-specific corrections (e.g., tips for squats, planks, etc.) and include step-by-step guides.
​
​
Creating an app
that helps people track their symptoms during
the Covid 19 pandemic.
Covid 19