This is the end of the case study!
Thank you for reading!
Always up for a coffee (no sweetener).
Let’s connect!
Leopard attacks on farms are rising due to habitat loss from deforestation, forcing them to seek prey on nearby cattle farms.
The Jio research team found that farmers were unable to respond quickly, losing cattle and other farm animals often.
Traditional solutions have too many limitations to solve this problem.
PROBLEM SPACE
Indian farmers are unable to protect themselves.
24H SURVEILLANCE?
Requires staff to observe cameras that farmers cannot afford.
LIVESTOCK ENCLOSURES?
Severely limits livestock to enclosed area - quality of life affected.
UX DESIGNER?
Fast, farmer retains control. But how tdo we implement this?
Alerting farmers through a smart camera + mobile application.
Preview of captured feed.
Allows farmers to accurately locate leopard sightings and take informed decisions.
Customizable alerts.
Offering farmers flexibility filters out non-essential information, reducing alert fatigue.
Support anytime.
SOLUTION
24/7 support through calling/text/chatbot integrated throughout application.
Mid-Fidelity Wireframes
For the rest, I had to build new screens, while ensuring I was following the JioGauSamriddhi design system.
For some screens, I had to take the existing fleshed out design, and ideated how to integrate the new feature.
Original Layout
Final Layout
Experiment 1
Alert takes priority on screen, focusing attention on the issue
Farmer would likely look at all cameras after alert to track movement - needs intuitive flow
Quick link to all cameras instead of going through “Devices” tab on top bar
Access to alert history, assisting in tracking if info was missed
Limited optionality on where to go from here
Alert takes priority on screen, focusing attention on the issue, but priority is given to tutorial for MVP
Experiment 1
Experiment 3
Final Mid-Fi Design
Experiment 2
Too little space for all details
Card-style interaction is unnecessary
Way to edit is not clear
Way to edit is not clear
Improve UX copy - unfamiliar language
Go back to prev. page to switch camera
Can switch between cameras without going back
Can identify location of camera using feed, without reading
Tags are part of design system, familiar and works with existing system
Main button is not prominent
Prioritizes feed first
Clear “Edit” button instead of card flipping
Prioritizes “Details” and “Alerts” over feed
Cannot identify camera without checking feed
MY CONTRIBUTION
I prioritized the 'Camera' page as the primary interface. All design decisions were research-backed while adhering to the established design system.
Leopard/Animal Detector Screen
Cameras
Video Feed
Homepage/Dashboard
Devices Screen
Explorations with various formats for the main screens. Although there were preset layouts in the application, I tried to keep my mind open to new ideas at this stage.
Exploring through sketches
DESIGN
A short impactful flow works best for our users.
My plan of action
1
Research
Desk Research
Competitive Analysis
Research Insights
User Journey
2
AI Model
Data Management
Model Adjustments
3
Design
Lo-Fidelity
Mid-Fidelity
Exported Components
4
Final Solution
Final Designs
Reflection
APPROACH
An end-to-end solution using AI to power a real-time alarm system, ensuring the safety of farmers and their livelihoods.
AI Alert System for Safety
Our AI model could detect danger with 92% accuracy.
Jio Platforms required all our MVP elements to be ready, so I used my programming background to create the AI model to detect dangerous animals - focusing on leopards.
Detecting Leopards!
Here are the results of the AI model, functioning on various stock videos that the model has not seen before.
AI MODEL
Some things we missed out on for this project were:
Usability testing
(the project was not given the budget, but let’s hope for the future!)Multiple languages on the interface
Evaluating interface with Harmonized Guidelines for Universal Accessibility in India
Reflecting on the experience.
I learned 2 key things through this project:
FARMING AND CULTURAL CONTEXT
Farming is an integral part of society, and the applications for the technology we contributed to can be seen in different cultural contexts. It would be interesting to design for American farmers in the future.
TIME BOUND PROJECTS
It was upsetting that usability testing was not completed before the end of the internship. However, this happens often in industry projects with low-budgets. I learned to adapt to the process that was feasible.
12 building blocks of a system for a new feature.
Here are some design components that were added to the Jio Design System after the project was completed.
The final flow was concise, quick, and insightful.
The flow would not be following the design goals if it were too long.
I minimized the number of clicks as far as possible, and optimized each step for user value and clarity.
Ability to move quickly from alert to monitoring to action.
Alerts displayed on home page for quick response.
Support integrated throughout application, over multiple platforms.
Contact Jio Support through various media.
Customizable alerts and notifications.
Change types of detection, sounds, and sensitivity.
A fast, intuitive, and supportive feature.
OUTCOMES
Users need quick alerts in a familiar system.
Lacking IoT experience, I reviewed academic literature for the system design. I developed an initial concept, which I later refined.
The IoT system design changed due to lack of data.
Initial Thoughts
AI Model
Phone App
IoT Device
Difficult for to create because would require large datasets of Indian animals.
Final System Design
Camera
Model
My design and dev area
User-Facing
AI Model
Phone App
Database
IoT Device
Solar Panel
Camera feed going through phone app gets reuploaded to cloud to populate the database.
RESEARCH
MY CONTRIBUTION
I saw the potential to repurpose captured feeds, using the images captured after the product was released to expand our Indian farm animal dataset.
Competitive analysis revealed that standard layouts work best for Indian users.
We examined farming security solutions and AI-based security solutions to inform the design of our application. We examined 8 solutions that were digital, physical, and some combination of both.
What were the user research takeaways?
I collaborated with the research team to understand the user pain points from the 7 qualitative unstructured interviews they had conducted on-site with Indian farmers.
Based on the research takeaways and user journey, we decided on a defined direction for our product.
Processes that take long not worth the work
Worry about support for tech issues
Fear of complications with setup
Worry about costing and maintenance
Minimal clicks from alert to action
24/7 tech support through phone calls
Video support for setting up system
MVP should be free, technicians on hand
Pain Points
Takeaways
Tenets and Design Goals
How do Indian farmers think about this problem?
To understand the mental model of the user group better, I constructed a user journey map. This also allowed me to focus on the exact opportunities needed for our solution.
Farmer receives early info about potential attack
Checks location of attack
Protects cattle and people
Is assured about effect
Things die down + adjusts app based on previous attack
Original user journey
Based on insights from farmers and research team
Proposed user journey (sketched)
Proposed user journey
Based on insights and prediction from research team
It is unlikely that this product will make this flow drastically easier for farmers. In the future, we would like to a solution that reduces the cognitive load further for our users.
CONTROL
Ability to update sensitivity of camera
Information about exact location of attack
SIMPLICITY
App is not overwhelmed with information
Limited capability, quick start-to-end flow
SUPPORT
Solve issues through multiple sources
Setting up made smooth by technicians
My role in this product was to complete a working prototype of the warning system, from leopard recognition to the design of the alarm feature in the mobile application.
I worked under the supervision of a design lead who guided me through every step of the process, especially the technicalities of dataset management and IoT integration.
JioGauSamriddhi is an information system that uses smart collars with a mobile application to monitor’s cattle behavior – heat trends, milk production, health – in an effort to make farms smarter.
Jio Platforms Ltd., is an Indian technology company with applications that span television, to cloud storage, to farming technologies.
What is the product?
What was my role?
Competitive Analysis
Journey Mapping
Data Cleaning + Augmentation
AI Engineering
Sketching
Wireframing
Prototyping
Product Design Internship
June 2023 - Nov 2023
Figma
FigJam
TensorFlow
Python
Google Colab
Roles & Responsibilities
Context
Tools
Individual + Manager
Team