Section 01
Problem Statement
No drone means no aerial photogrammetry? Wrong — a phone video walked around a site or stairwell gives a dense camera path; the phone's gravity/linear-acceleration sensors plus ARCore provide per-frame pose. A lightweight structure-from-motion result can be pre-generated on a laptop or free Colab server, but the defect-detection model and reporting must run on-device so a worker standing by the slab can tap the screen and see cracks immediately. The team captures the actual video with a phone and the phone is the worker's capture and detection interface — a real drone is not allowed.
Section 02
Solution Requirements
Phone-video capture protocol
Record a 40–60s video walking around the structure with camera + GPS metadata + ARCore pose stored per keyframe (CSV/SQLite); at least 110 geotagged key-frames extracted, fully on the phone.
BIM / 3D reference on phone
Load a simple pre-converted .glb/.obj into a three.js-in-WebView viewer that rotates on the phone.
On-device defect detection
A TFLite model (EfficientNet-seg or YOLOv8n) running on the phone classifies 3 defect classes: concrete crack (measured in mm), rebar spacing (mm), and honeycombing/missing cover — drawing boxes/masks with a severity label from an on-screen ruler, on live phone-captured frames.
Deviation vs. design (phone-side check)
Selecting an element (e.g. "column 5B") loads the designed rebar spacing from a JSON DB, compares it to the live-detected spacing, and outputs the error in mm with pass/warn/fail color-coding alongside the real photo.
Automated QA PDF report on phone
Tapping "Generate report" (PdfDocument / PrintManager) produces a PDF with date, GPS location, defect count, severity table, defect photo thumbnails, and a QR stamp containing project + timestamp — all generated on-phone.
Ticketing (phone to mock API)
Each failure posts a simulated ticket over Retrofit + OkHttp to a small Firebase/Supabase/mock server; a supervisor sees it via a web panel, and the creation is triggered from the phone.
Offline defect model
The defect model runs without network; in airplane mode the defect report is still produced offline, and the ticket syncs once connectivity returns.
Section 03
Expected Solution / Outcomes
- Video opens with a phone recording a real walk-around of a cracked wall or mis-aligned formwork.
- Live TFLite bounding box shows e.g. "1.2mm crack" on the phone screen inside a real room/gallery.
- Selecting an element in the 3D/AR viewer shows e.g. "error 14mm".
- A PDF export opens inside a PDF viewer on the same phone.
- A second reviewer sees the uploaded ticket.
Section 04
Technology & Hardware Constraints
Data: Mobile crack dataset (Kaggle), 3D reference .obj, designed rebar spacing JSON.
Section 05
Scoring
Requirements Completeness
50
Each requirement (R-id) is worth 3–12 marks. A hard-coded or mocked claim earns 0 — only working code counts.
Presentation & Demo
50
3-slide pitch + video ≤ 7 min + live demo. Judges must understand the problem in ≤ 90 seconds.
Real-world Implementation & Feasibility
50
Would a PM, engineer, or worker install this with only a phone on Day 1 of the next project? Must work offline and cheaply. Field-trial videos prove real-environment execution (20+ pts of this score is the field video).
Innovation & Disruption
50
A novel fusion of phone sensors (e.g. IMU + camera + barometer = scaffold-instability detector) is rewarded over simple API-calling. Re-skinning an existing SaaS scores low.
Section 06
Bonus (+10)
"Continuous scan mode": swipe the phone slowly over the floor like a scanner; on-device pose + depth (ARCore Depth API on supported devices) builds a per-square-metre severity heat-map, exportable as a severity CSV.
Section 07
Deliverables
| Item | Format |
|---|---|
| Video Demo | ≤ 7 min, YouTube/Vimeo unlisted link. Must open with the phone alone on a table (no external hardware). |
| Field Trial Videos | (a) Walking a phone around the outside of a real building to produce a defect video. (b) Inspecting a real crack/misalignment/mis-placed rebar with the phone detecting it live. |
| Demo Link | Android APK download link (installed on the reviewer's own phone) or a mobile-optimised web app URL. |
| GitHub | capture_app/, defect_model.tflite, compare_engine/, pdf/, ticketing/, README explaining the PC-side 3D pipeline and noting the phone is the operator. |
| Pitch Deck | 3-slide PDF (Problem → Solution → Field-Proof & Impact). |
| Technical Brief | 2-page PDF: architecture diagram, phone-sensor map, tech stack, limitations, PRIVACY.md. |
Section 08
Field-Trial Requirements
- Real outdoor walk-around video producing a defect capture.
- Live on-phone detection of a real crack, misalignment, or mis-placed rebar.
Section 09
Privacy Requirements
Any human face captured in a field video must be consented or blurred. A PRIVACY.md file is required in every repository.