StreamPETR with EVA-02 Backbone

This model card describes a StreamPETR-style camera-only 3D object detection configuration using an EVA-02-Large image backbone for nuScenes.

No model weights are included in this repository.

Model Details

Architecture: StreamPETR-style temporal 3D detector
Backbone: EVA-02 Large ViT, patch size 16, embed dim 1024, depth 24, 16 heads
Input: Multi-view camera images
Dataset: nuScenes
Task: Camera-only 3D object detection
Framework: MMDetection3D / StreamPETR custom plugin

The local implementation also includes RepDETR3D-compatible EVA-02 configs. This card is intentionally metadata-only and does not publish checkpoints.

Intended Use

Use this repository as a lightweight public model entry for the StreamPETR EVA-02 configuration. To run or train the model, use the corresponding StreamPETR codebase and local configs/checkpoints.

Weights

Weights are not uploaded here. Users must provide their own pretrained EVA-02 and detector checkpoints.

Citation

If you use StreamPETR, cite the original StreamPETR work and any EVA/EVA-02 backbone references used by your experiment.

Downloads last month: -; Downloads are not tracked for this model. How to track