CoCare AI – The Respectful Copilot for Inclusive Care

🌟 Inspiration

Our team was inspired by organizations like Mariam’s Footsteps, who work with children with disabilities and orphans, many of whom struggle to express their needs and feelings. We saw the emotional labor caregivers undertake daily to observe, log, and personalize care, often without knowing how a child truly feels. We wanted to build something that respects every child’s dignity while easing caregiver burnout.

🤖 What it does

CoCare AI is a respectful AI copilot for inclusive care. It uses privacy-safe edge sensors to detect activities like play, rest, and agitation (without storing raw video or audio), then generates smart summaries every 30 to 60 minutes for caregivers to quickly edit or approve.
It empowers children to express themselves using tablet buttons or eye-tracking inputs, folding their voice directly into care logs. Additionally, it offers calming sensory games that help relax and engage the child while gently monitoring their emotional state.

🛠️ How we built it

We built CoCare AI using:

• Streamlit for rapid, accessible UI prototyping
  
• TensorFlow Lite for on-device activity recognition
  
• A privacy-first edge processing pipeline to ensure compliance and safety
  

We prioritized real-time performance, accessibility, and ethical design throughout the process.

🚧 Challenges we ran into

• Balancing real-time detection accuracy with low computational load on edge devices
  
• Designing UIs accessible to both caregivers and children with different abilities
  
• Seamlessly integrating sensory games while maintaining monitoring accuracy
  
• Ensuring privacy and trust when working with vulnerable populations

🏆 Accomplishments that we're proud of

• Built a functioning prototype with edge-based pose detection and activity summarization
  
• Designed a voice-enabling UI for children with limited verbal communication
  
• Maintained privacy-by-design in every system component
  
• Developed a calming game interface that simultaneously monitors and soothes
  

📚 What we learned

• Privacy is essential when working with vulnerable populations
  
• Caregivers need editable summaries, not just raw AI insights
  
• Even small features like a simple "I liked this" button can make a child feel heard and empowered
  
• Inclusive design means designing with, not just for, caregivers and children

🔮 What's next for CoCare AI – The Respectful Copilot for Inclusive Care

• Integrate emotional state detection for deeper understanding
  
• Collaborate with organizations like Mariam’s Footsteps to tailor CoCare AI to real-world settings
  
• Expand our sensory game library to include adaptive educational tools
  
• Further refine the UI/UX for broader accessibility, including gesture- and gaze-based controls
  

We built CoCare AI to give every child a voice and every caregiver peace of mind.

Built With

• mediapipe
• opencv
• python
• react
• streamlit
• tensorflowlite
• typescript