Thesis
Smart AAC System with Facial Expression Recognition for Autism
- Author
- Yasas Pasindu Fernando
- Supervisors
- Ms. Niruni Fonseka; Mr. Akila Udara Akalanka
- Module
- CS6P05ES Final Report
- Date
- 22 May 2026
Research
Smart AAC research
Trilingual offline AAC with on-device facial expression recognition for autism support in Sri Lanka
Final-year Software Engineering project with a conference paper track. The goal was affordable AAC that supports Sinhala, Tamil, and English, runs offline, and keeps facial-expression inference on the device.
Conference manuscript
Trilingual Offline Smart AAC with On-Device Facial Expression Recognition for Autism in Sri Lanka
- Venue
- EICON 2026, ESOFT International Conference
- Paper ID
- FPC21
- Authors
- Fernando E. Y. P., Fonseka N., Akalanka P. D. A. U.
- Status
- Full paper submitted; major revision completed; camera-ready manuscript prepared (July 2026). Submission is not the same as acceptance or published proceedings unless later confirmed.
Key findings (as reported)
- Prototype checked with unit tests, TFLite tensor-contract checks, real-device walkthroughs, and Google Play internal testing.
- Primary FER model test accuracy 52.9% (weighted F1 0.54) on a public-dataset held-out split. Useful only as a cautious supportive cue under caregiver supervision.
- Anonymous early tester questionnaire (n=11): 9 of 11 thought the app could help a child express basic needs. No children were recruited; no clinical pilot was run.
Boundaries
- • Not a diagnostic medical device.
- • Not Ministry of Health approved for clinical use.
- • Pilot intended only after ethical clearance (Pragathi Centre / National Hospital Galle letter of support).
- • No claim of production healthcare deployment.