Commit 859e9c
2026-08-07 11:23:58 Anonymous: Product Overview| /dev/null .. engineering roadmap.md | |
| @@ 0,0 1,75 @@ | |
| + | --- |
| + | title: Product Overview |
| + | aliases: [3netra Product, Mission] |
| + | tags: [3netra/wiki, product, personas] |
| + | product: 3netra Meta Android App |
| + | source_repository: 3Netra-ai/3n_meta_android |
| + | source_ref: origin/prod |
| + | source_revision: b48aada8fd5427b5db0c02945569963c4ae543c8 |
| + | last_verified: 2026-08-06 |
| + | up: "[[Home]]" |
| + | --- |
| + | |
| + | # Product Overview |
| + | |
| + | ## Mission |
| + | |
| + | **3netra** (Sanskrit for "third eye") gives blind and visually impaired higher education students (high school and college) real-time, hands-free audio awareness of their academic world: reading slides, whiteboards, and textbooks, narrating campus scenes, recognizing classmates and professors, detecting emotions — through ordinary-looking Meta Ray-Ban glasses. |
| + | |
| + | > [!quote] Core promise |
| + | > "Every student deserves to explore, learn, and belong — 3netra makes that possible." |
| + | |
| + | ## Problem |
| + | |
| + | ~285M people worldwide live with visual impairment (39M blind). In higher education (high schools and universities), VI students can't read slides/whiteboards, miss non-verbal social cues, and struggle in dynamic lecture halls, labs, and campus grounds. Existing tools (canes, Braille, screen readers) don't provide real-time environmental awareness, and no product unifies text reading + scene understanding + emotion detection + face ID in one accessible wearable. |
| + | |
| + | ## Goals & KPIs (PRD v1.0) |
| + | |
| + | | Metric | Target | |
| + | |---|---| |
| + | | Scene description latency (capture→audio) | < 2 s | |
| + | | OCR accuracy | > 95% | |
| + | | Emotion detection accuracy | > 85% | |
| + | | Face recognition precision | > 90% | |
| + | | CSAT | > 4.5/5 | |
| + | | DAU/MAU | > 60% | |
| + | | Avg session length | > 15 min | |
| + | | Onboarding completion | > 80% | |
| + | |
| + | ## Personas |
| + | |
| + | ### ⭐ Aiden — The Curious Student (primary) |
| + | |
| + | 19 y/o, college sophomore, congenital blindness, university campus. Can't read lecture slides/whiteboards clearly, misses who's speaking in seminars, struggles in crowded lecture halls and campus grounds. Wants: read slides, identify professors and classmates, navigate campus independently, follow presentation diagrams. Delight: 3netra identifies lab equipment — "I see a compound light microscope on your desk!" |
| + | |
| + | ### Neha — The High School Senior |
| + | |
| + | 17 y/o, high school senior preparing for college, low vision (light/shapes only). Reads handouts/textbooks, deciphers lab diagrams, reads instructor expressions, navigates the university library. Self-conscious about assistive tech — needs glasses that look normal. |
| + | |
| + | ### Mrs.Adhikary — Educator & Disability Services Coordinator |
| + | |
| + | 38, University Disability Services Specialist / High School Accessibility Coordinator. Monitors usage, customizes vocabulary/persona, reviews academic summaries, ensures student-centered accessibility. |
| + | |
| + | ## Implementation boundary |
| + | |
| + | The reviewed Android module targets API 29–34 and initializes Meta Wearables DAT 0.7.0. It contains camera-capture and registration code, but no hardware run at this revision proves a supported glasses model, capture latency, gesture delivery, or speaker routing. Hardware specifications and product latency targets above are PRD/document claims, not verified release behavior. |
| + | |
| + | The current Compose implementation still contains legacy K–6 labels and child-voice filtering. The higher-education positioning on this page is editorial/product direction; implementation alignment remains an open decision. |
| + | |
| + | ## Interaction Model |
| + | |
| + | - **Glasses gestures:** `BluetoothService` has callable scene/read/long-press/emotion handlers, but the reviewed source does not register DAT gesture callbacks. The Home scene action invokes the single-tap handler directly; do not treat the gesture map as hardware-verified. |
| + | - **Voice:** `VoiceCommandService` can run Android speech recognition after `onLongPress`; no wake-word detector or verified glasses-mic path is present. |
| + | - **Phone UI:** big-tap-target Compose screens with 4-tab bottom nav. See [[03-Frontend-Navigation]]. |
| + | |
| + | ## Compliance Posture |
| + | |
| + | FERPA/COPPA and privacy statements are requirements, not demonstrated compliance. The active enrollment/matching path reads and writes face embeddings through Supabase; the encrypted local utility is unreferenced. See [[08-Data-Model-Supabase]] and [[11-Security-Privacy-Compliance]]. |
| + | |
| + | > [!info] Audit provenance |
| + | > Implementation claims on this page were reviewed at `3Netra-ai/3n_meta_android`, `origin/prod` commit `b48aada8fd5427b5db0c02945569963c4ae543c8` (2026-08-05). The local checkout directory is named `archived-3n-meta-app`. |
| + | |
| + | --- |
| + | |
| + | > [!tip] Navigation |
| + | > ⬅️ [[Home]] · ➡️ [[02-System-Architecture|System Architecture]] |