Section 01
Problem Statement
A project manager in the field needs to know why the slab is late in 5 minutes, not in a 2-week report. The phone holds the weekly manual progress entry (photo + voice note + numbers), the weather comes from GPS + a free API, the material price index comes from cached app data, and the ML model runs on the phone in TFLite (a small XGBoost model converted) or a serverless endpoint. The phone's camera is a field progress-capture tool: photograph the slab, get an on-device classification of formwork vs. reinforcement vs. poured, feeding the variance forecast.
Section 02
Solution Requirements
Historical-project data loaded
The app bundles 400+ historical projects (CSV → Room DB) with schedule/cost columns.
Field progress capture (photo, voice)
A real phone photo of a slab zone is classified by an on-device EfficientNet-Lite model ("not started / formwork / rebar / poured"), or a voice note via speech-to-text extracts a phrase like "rebar 70%" — this is the live feature update.
Forecast model on the phone
A converted XGBoost model (via treelite/ONNX→TFLite) predicts schedule slip (days) and cost variance (%) locally in < 0.2s, e.g. "you will be 6 days late, 1.2% over budget."
Explainable AI drivers
For every prediction, list the top-3 drivers from a pre-generated SHAP-style lookup or on-device counterfactual, e.g. "Change order C2 (+3d), 5 rainy days 14mm (+2d)", shown on a dashboard bar on the phone.
Scenario slider
A SeekBar slider for "add crew / cement price / add 2 rainy days" recomputes the live forecast on the same screen instantly.
Recommendations
For a bad forecast, list 3 rule-based recommendations with their expected delta (e.g. "use admixture → -2d, -3k") and a "send to PM" share intent.
Dashboard view
A budget S-curve / per-activity Gantt drawn via MPAndroidChart / Victory in WebView on the worker's phone.
Offline operation
Model, database, and historical data work without network except for optional live weather.
Section 03
Expected Solution / Outcomes
- A worker on a balcony/porch takes a phone photo → screen shows "rebar" classification → progress log updates.
- The dashboard shows "you'll be 5 days late".
- Tapping a driver highlights "Change order C2".
- Dragging a slider to "add 2 rainy days" moves the prediction to 7.5 days.
- Recommendations appear and the demo runs in airplane mode.
Section 04
Technology & Hardware Constraints
Data: Bundled historical project CSVs.
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)
A Monte-Carlo P50/P80/P95 distribution with a progress bar, or auto-extracting change orders from 500 text sentences using speech-to-text + an on-device TFLite text classifier for delay-risk.
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) Real photo progress capture on an actual construction/porch/stairwell. (b) A real city walk showing an outdoor temperature reading (weather widget/API) + a change-order voice note, plus a slider demo. |
| Demo Link | Android APK download link (installed on the reviewer's own phone) or a mobile-optimised web app URL. |
| GitHub | data/, training/, tflite_model/, xai/, forecast/, dashboard/. |
| 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 photo-based progress capture on a real construction/porch/stairwell.
- Real outdoor weather + voice change-order note, with a live slider demo.
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.