MCG-NJU/SparseBEV
A sparse transformer model for 3D object detection from multi-camera video feeds, designed for autonomous driving perception.
Not currently ranked — collecting fresh signals.
star history
SparseBEV is a PyTorch implementation of a high-performance 3D object detection system that uses sparse sampling and transformer architectures to process multi-camera video data. The model converts 2D camera inputs into 3D bird’s-eye-view representations to detect objects in autonomous driving scenarios. It was published at ICCV 2023 and extended to TPAMI 2026.
Frequently asked
- What is MCG-NJU/SparseBEV?
- A sparse transformer model for 3D object detection from multi-camera video feeds, designed for autonomous driving perception.
- Is SparseBEV open source?
- Yes — MCG-NJU/SparseBEV is open source, released under the MIT license.
- What language is SparseBEV written in?
- MCG-NJU/SparseBEV is primarily written in Python.
- How popular is SparseBEV?
- MCG-NJU/SparseBEV has 465 stars on GitHub.
- Where can I find SparseBEV?
- MCG-NJU/SparseBEV is on GitHub at https://github.com/MCG-NJU/SparseBEV.