The first Brazilian traffic dataset for autonomous AI.
The world’s autonomous driving models were trained on data from San Francisco, Phoenix, Shanghai, and Seoul. No public or private dataset covers Brazilian traffic with sufficient quality to train an autonomous system.
The Bex Dataset changes that.
What it contains
- Video — hours of continuous recording of Brazilian traffic, captured by the Bex Cam.
- Telemetry — speed, acceleration, braking, vehicle OBD data.
- Location — high-precision GPS, synchronized with video.
- Annotations — identified objects, traffic events, road conditions.
Quality
The dataset’s quality determines the quality of everything that follows. Every frame is curated, annotated, and validated. Noisy, incomplete, or biased data is discarded. The goal is not volume — it is representativeness.
Licensing
The Bex Dataset is proprietary. Available for licensing by automakers, research centers, and navigation companies that need to train or validate models for the Brazilian context.