# Faster Training, Fewer Labels: Self-Supervised Pretraining for Fine-Grained BEV Segmentation

arXiv: https://arxiv.org/abs/2602.18066 · HTML: https://mlsanity.com/paper/2602.18066/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 28 of its 34 references that are in the map.
- First posted on arXiv 20 Feb 2026.

## Builds on (most cited first, 28 of 28)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [nuScenes: A multimodal dataset for autonomous driving](https://mlsanity.com/paper/1903.11027/) (2019)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [Scalability in Perception for Autonomous Driving: Waymo Open Dataset](https://mlsanity.com/paper/1912.04838/) (2019)
- [Masked-attention Mask Transformer for Universal Image Segmentation](https://mlsanity.com/paper/2112.01527/) (2021)
- [BEVFormer: Learning Bird's-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers](https://mlsanity.com/paper/2203.17270/) (2022)
- [Lift, Splat, Shoot: Encoding Images From Arbitrary Camera Rigs by Implicitly Unprojecting to 3D](https://mlsanity.com/paper/2008.05711/) (2020)
- [BEVFusion: Multi-Task Multi-Sensor Fusion with Unified Bird's-Eye View Representation](https://mlsanity.com/paper/2205.13542/) (2022)
- [Accelerating 3D Deep Learning with PyTorch3D](https://mlsanity.com/paper/2007.08501/) (2020)
- [KITTI-360: A Novel Dataset and Benchmarks for Urban Scene Understanding in 2D and 3D](https://mlsanity.com/paper/2109.13410/) (2021)
- [BEVFormer v2: Adapting Modern Image Backbones to Bird's-Eye-View Recognition via Perspective Supervision](https://mlsanity.com/paper/2211.10439/) (2022)
- [Panoptic nuScenes: A Large-Scale Benchmark for LiDAR Panoptic Segmentation and Tracking](https://mlsanity.com/paper/2109.03805/) (2021)
- [M2BEV: Multi-Camera Joint 3D Detection and Segmentation with Unified Birds-Eye View Representation](https://mlsanity.com/paper/2204.05088/) (2022)
- [Simple-BEV: What Really Matters for Multi-Sensor BEV Perception?](https://mlsanity.com/paper/2206.07959/) (2022)
- [FB-BEV: BEV Representation from Forward-Backward View Transformations](https://mlsanity.com/paper/2308.02236/) (2023)
- [Bird's-Eye-View Panoptic Segmentation Using Monocular Frontal View Images](https://mlsanity.com/paper/2108.03227/) (2021)
- [SkyEye: Self-Supervised Bird's-Eye-View Semantic Mapping Using Monocular Frontal View Images](https://mlsanity.com/paper/2302.04233/) (2023)
- [S4C: Self-Supervised Semantic Scene Completion with Neural Fields](https://mlsanity.com/paper/2310.07522/) (2023)
- [Mask2Map: Vectorized HD Map Construction Using Bird's Eye View Segmentation Masks](https://mlsanity.com/paper/2407.13517/) (2024)
- [PointBeV: A Sparse Approach to BeV Predictions](https://mlsanity.com/paper/2312.00703/) (2023)
- [360BEV: Panoramic Semantic Mapping for Indoor Bird's-Eye View](https://mlsanity.com/paper/2303.11910/) (2023)
- [DifFUSER: Diffusion Model for Robust Multi-Sensor Fusion in 3D Object Detection and BEV Segmentation](https://mlsanity.com/paper/2404.04629/) (2024)
- [High-Definition Map Generation Technologies For Autonomous Driving](https://mlsanity.com/paper/2206.05400/) (2022)
- [DualBEV: Unifying Dual View Transformation with Probabilistic Correspondences](https://mlsanity.com/paper/2403.05402/) (2024)
- [RendBEV: Semantic Novel View Synthesis for Self-Supervised Bird's Eye View Segmentation](https://mlsanity.com/paper/2502.14792/) (2025)
- [GLane3D : Detecting Lanes with Graph of 3D Keypoints](https://mlsanity.com/paper/2503.23882/) (2025)
- [BEVCon: Advancing Bird's Eye View Perception with Contrastive Learning](https://mlsanity.com/paper/2508.04702/) (2025)

## Abstract (arXiv)

Dense Bird's Eye View (BEV) semantic maps are central to autonomous driving, yet current multi-camera methods depend on costly, inconsistently annotated BEV ground truth. We address this limitation with a two-phase training strategy for fine-grained road marking segmentation that removes full supervision during pretraining and halves the amount of training data during fine-tuning while still outperforming the comparable supervised baseline model. During the self-supervised pretraining, BEVFormer predictions are differentiably reprojected into the image plane and trained against multi-view semantic pseudo-labels generated by the widely used semantic segmentation model Mask2Former. A temporal loss encourages consistency across frames. The subsequent supervised fine-tuning phase requires only 50% of the dataset and significantly less training time. With our method, the fine-tuning benefits from rich priors learned during pretraining boosting the performance and BEV segmentation quality (up to +2.5pp mIoU over the fully supervised baseline) on nuScenes. It simultaneously halves the usage of annotation data and reduces total training time by up to two thirds. The results demonstrate that differentiable reprojection plus camera perspective pseudo labels yields transferable BEV features and a scalable path toward reduced-label autonomous perception.