ROAD SECTIONS ▤
—Sections in latest analysis SYSTEM WORKSPACE
ADVANCED PAVEMENT ANALYTICS
Understand your roads.
Prioritize what matters.
A data-driven workflow for pavement condition assessment, distress intelligence and maintenance investment planning.
Python-powered · GIS-enabled · AI-ready
01ROAD
INTELLIGENCE
INTELLIGENCE
23°48' N
90°24' E
90°24' E
AVERAGE CONDITION ◉
—/100Network condition indexHIGHEST PRIORITY ⚑
—Lowest condition scoreAI MODEL STATUS ◎
Not configuredCustom model requiredENGINEERING PERFORMANCE
Condition intelligence
Condition score by sectionScale: 0–100 · Higher is better
● Condition scoreCondition distributionRoad sections by category
0SECTIONS
DECISION SUPPORT
Maintenance priority ranking
| RANK | SECTION / ROAD | CONDITION | CATEGORY | RECOMMENDED ACTION |
|---|---|---|---|---|
| Upload your project CSV to view road rankings. | ||||
SPATIAL INTELLIGENCE
GIS road network
● Good● Fair● Poor● CriticalNo road geometry loaded
COMPUTER VISION · FREE PUBLIC MODEL
AI pavement distress detection
◎
Road Hazard Detection — YOLO11n
Use the public browser-based detector below. It runs the model on your device with ONNX and WebAssembly; no RoadVision Pro AI server, Hugging Face subscription or API key is required.
PotholeLongitudinal crackTransverse crackAlligator crack
Open detector in a new tab ↗Public demo by Vivek Vagale
If the detector does not appear inside this panel, use “Open detector in a new tab”. The public demo is hosted and maintained by its author; availability and model downloads depend on that service.
ENGINEERING SAFEGUARDS & MODEL LIMITATIONS
AI output is a preliminary screening result, not an engineering measurement.
- Detected damage must be checked by a human engineer.
- The model was trained on Indian RDD2022 road images, not specifically on Bangladeshi roads.
- Published held-out test mAP@0.5 is approximately 0.386; transverse-crack detection is particularly weak.
- Do not convert image pixels into crack area, pothole depth or maintenance quantities without field measurement and calibration.
- AI results do not automatically change RoadVision Pro's original pavement condition score.
Model source and evaluation: VivekVagale/road-hazard on GitHub.
CAPITAL PLANNING
Maintenance budget optimization
Enter a documented cost and benefit value for each road. Benefit scores are supplied by the project team—not generated arbitrarily.
Optimization results will appear here after you provide road costs and benefit values.
DATA PIPELINE
Import & validate dataset
Upload pavement survey data
Use your original road_pavement_raw_data.csv file or another CSV with the required engineering columns.
Required fields
Section_ID, Road_Name, Section_Length_m, Pavement_Width_m, Crack_Percent, Pothole_Area_m2, Rut_Depth_mm, Roughness_Rating, Drainage_RatingWaiting for dataset
↓ Download analyzed CSVPROJECT DELIVERABLES
Reports & submission center
Google Drive submission links can be added in frontend/app.js after the team creates the final shared files.
THE PEOPLE BEHIND THE PROJECT
Project team
SERIAL 01 · 24205135
Rakibul Hasan
Group Leader · Developer
SERIAL 03 · 24205106
Md. Mahedi Morshed
Regular Member
SERIAL 02 · 24205108
Niloy
Regular Member
SERIAL 04 · 24205123
Rafiul Islam
Regular Member
Designed & Developed by Rakibul Hasan · Foundation X
ENGINEERING TRANSPARENCY
Methodology & project information
Weighted pavement condition model
Cracking 25% · Potholes 25% · Rutting 20% · Roughness 20% · Drainage 10%
Condition Score = 100 − Weighted Distress Score. The original Python engineering engine remains the single source of truth.Classification thresholds
Good ≥ 85 · Fair ≥ 70 · Poor ≥ 40 · Critical < 40
AI screening is experimental and does not automatically modify engineering scores.