Datasets:

origindatalab
/

bangladesh-urban-traffic-free-sample-v2

Tasks:

Object Detection

Modalities:

Formats:

Languages:

Size:

Tags:

Libraries:

License:

Dataset card Data Studio Files Files and versions

xet

Community

Dataset Viewer

Auto-converted to Parquet Duplicate

Split (1)

train · 9 rows

Search is not available for this dataset

image imagewidth (px) 1.28k 1.44k

Bangladesh Urban Traffic Dataset (Dhaka Mixed Traffic Dataset) — Free Sample

This Bangladesh Traffic Dataset contains real-world mixed traffic scenes including rickshaws, motorcycles, buses, cars, pedestrians, market congestion, intersections, and complex urban mobility patterns.

The dataset is designed for computer vision, object detection, autonomous driving, ADAS, robotics, perception systems, physical AI, traffic analytics, and emerging-market mobility research.

This repository serves as a free sample of a larger Dhaka Traffic Dataset and Mixed Traffic Dataset developed by Origin Data Lab.

Commercial Access Available

Need a larger mixed-traffic dataset for model evaluation or internal PoC?

Discovery Pack: $500 — 120 curated 15-second clips
Evaluation Pack: $2,000 — broader enterprise evaluation dataset
Use cases: Autonomous Driving, Robotics, ADAS, Physical AI, Object Detection
Delivery: secure commercial download after checkout

👉 Buy Discovery Pack: https://origindatalab.lemonsqueezy.com/checkout/buy/46ece78e-7e3e-4c5f-87ca-95f582934d0a

Website: https://origindatalab.io
Email: [email protected]

Gallery

Representative scenes from Dhaka, Bangladesh, including flooded roads, market zones, rickshaw-dense environments, pedestrian-heavy interactions, mixed traffic corridors, and low-light urban mobility scenarios.

These environments represent real-world edge cases that are rarely covered by conventional Western driving datasets.

Flooded Mixed Traffic Waterlogged road with buses, rickshaws, and pedestrians.	Rickshaw-Dense Corridor High-density informal urban mobility environment.	Low-Light Urban Mobility Dusk conditions with mixed traffic and reduced visibility.
Flooded Traffic Interaction Multi-class traffic operating under flood conditions.	Market Zone Pedestrian Density Dense pedestrian activity near informal roadside markets.

Preview Privacy Notice

Preview images shown in this repository use aggressive masking to accelerate public sharing and privacy protection.

Commercial delivery packages include production-grade privacy processing workflows designed to preserve scene structure, object geometry, and computer vision usability while protecting personal identity information.

Preview Video

This short preview demonstrates real-world mixed traffic conditions collected in Dhaka, Bangladesh, including:

Rickshaws
Motorcycles
Buses
Pedestrians
Market-zone congestion
Dense urban interactions

Watch the Video

▶ Watch the Bangladesh Urban Traffic Preview Video

Bangladesh Urban Traffic Preview Video

If the embedded player is supported by your browser, the video can be viewed directly from this repository.

Annotation Preview

Representative annotation previews generated from the Origin Data Lab labeling pipeline.

Note: Preview images are privacy-reduced examples. Commercial deliveries include additional review assets and metadata under applicable licensing terms.

Classes currently demonstrated:

person
car
bus
truck
motorcycle
bicycle
rickshaw

Commercial Access

This repository is a free technical preview intended for buyer evaluation and technical review.

Full commercial delivery is available separately.

Discovery Pack purchases can be completed instantly through secure checkout.

Discovery Pack — $500

Designed for initial evaluation and proof-of-concept projects.

👉 Buy Discovery Pack ($500)

Instant commercial access after checkout.

Includes:

120 curated traffic clips
Mixed traffic scenarios
Market-zone traffic
Intersection traffic
Dense pedestrian interactions
Metadata package
Commercial evaluation license

Evaluation Pack — $2,000

Designed for enterprise evaluation, internal model testing, and broader mixed-traffic robustness review before a larger commercial engagement.

Includes:

Extended evaluation dataset
Broader scene coverage than the Discovery Pack
Additional metadata coverage
Enterprise PoC support
Buyer consultation support

Contact

Website: https://origindatalab.io

Email: [email protected]

Search Keywords

This dataset may also be relevant for users searching for:

Bangladesh Traffic Dataset
Dhaka Traffic Dataset
Urban Traffic Dataset
Mixed Traffic Dataset
Rickshaw Dataset
Road Scene Dataset
Autonomous Driving Dataset
ADAS Dataset
Computer Vision Dataset
Physical AI Dataset
Robotics Dataset
Traffic Video Dataset

Why This Dataset Matters

Most autonomous driving datasets focus on structured road environments.

Dhaka traffic presents a fundamentally different challenge:

Rickshaws
CNG auto-rickshaws
Informal lane behavior
Dense pedestrian interactions
Market congestion
Occlusions
Unstructured vehicle movement
Mixed mobility environments

These characteristics create valuable edge cases for:

Autonomous driving
ADAS
Object detection
Multi-object tracking
Robotics
Physical AI
Domain adaptation
Annotation planning
Emerging-market mobility research

What Is Included In This Repository

This public repository contains:

Preview video
buyer_sample_index.csv
dataset_manifest.json
Hero image
Gallery images
Annotation preview
Dataset card

This repository is intended for technical review and dataset evaluation.

The complete commercial dataset is delivered separately.

Dataset Statistics & Technical Overview

This commercial dataset is derived from a larger Bangladesh urban mobility inventory built for robotics, perception, autonomous driving, ADAS, and Physical AI evaluation.

Inventory Scale

39,041 analyzed traffic segments
5,921 unique source videos
39,041 buyer-value scored segments
12,148 GPS-enriched segments
12,710 segments with object-count metadata
11,922 estimated car instances across object-count-ready segments
Privacy-reduced preview assets available

The Discovery Pack is a curated commercial subset selected from this larger analyzed inventory.

Representative Scene Distribution

Scene Type	Segment Count
Intersection Traffic	10,182
Night / Low-Light Mobility	5,635
Market Areas	2,754
Mixed Traffic Roads	2,189
Vehicle-Rich Corridors	1,381
Tracking / Occlusion Scenarios	1,110
Straight Urban Roads	448
Pedestrian-Dense Areas	214

Buyer-Facing Quality Distribution

Quality Tier	Segment Count
Strong Commercial Candidates	10,330
Usable Commercial Candidates	11,792
Commercially Usable Total	22,122

Sensor Metadata Distribution

Metadata Tier	Segment Count
Prime Sensor Metadata	8,086
Standard Sensor Metadata	20,425
GPS-Enriched Segments	12,148
Object-Count Metadata Available	12,710

Representative Object Categories

Object categories represented throughout the collection include:

Rickshaw
CNG Auto-Rickshaw
Motorcycle
Pedestrian
Car
Bus
Truck
Bicycle

Metadata Signals

Commercial packages may include:

GPS metadata
IMU metadata
Motion-related metadata
Scene classification tags
Traffic density indicators
Object-count metadata
Privacy QA metadata
Buyer-facing evaluation indexes

Preview Summary

Country: Bangladesh
City: Dhaka
Scenario Type: Urban Mixed Traffic
Data Type: Video Preview + Metadata Sample
Commercial Pack Available: Yes
Discovery Pack Size: 120 Curated Clips
Delivery Format: Individual 15-second clips
Commercial Delivery: Secure download link after purchase or direct buyer handoff

Additional commercial metadata includes:

GPS metadata
IMU metadata
Scene labels
Object counts
Privacy QA reports
Buyer-facing indexes

Metadata Overview

Commercial packages include metadata designed for dataset filtering and technical review.

Examples include:

Object composition metadata
GPS motion metadata
IMU motion metadata
Lighting metadata
Traffic density metadata
Tracking difficulty metadata
Annotation readiness metadata
Privacy QA metadata
Buyer-use-case tags

The files in this repository provide a sample of the metadata structure.

Transparency Notice

Origin Data Lab follows a transparency-first policy.

Unless explicitly marked as human-verified:

Object labels may be machine-generated weak labels.
Tracking metrics may be estimated.
Environmental metadata may contain automated enrichments.
GPS and IMU signals should be independently reviewed for safety-critical applications.

Recommended uses: