ModelPrésentéExterne

Beval Model Zoo

This repository contains the pre-trained model weights from the paper:

Sourcehuggingface.co/manutheeng/beval_model_zoo/tree/40006d03b5e55bf4d428cd1c03419af2776dd91b

TypeOtherVersionRevision 40006d0CClasse

Source du modèle

Type: Other
Taille du fichier: 54,8–66,5 MB
Version: vRevision 40006d0
Mis à jour: 22 oct. 2026
Téléchargements: 0
Source: Hugging Face

Description de la source

BEVal Model Zoo — BEV Segmentation Models for Autonomous Driving

This repository contains the pre-trained model weights from the paper:

BEVal: A Cross-dataset Evaluation Study of BEV Segmentation Models for Autonomous Driving
Manuel Diaz-Zapata, Wenqian Liu, Robin Baruffa, Christian Laugier
2024 18th International Conference on Control, Automation, Robotics and Vision (ICARCV), pp. 704–709, IEEE.
arXiv GitHub

Overview

This collection contains 60 pre-trained models for semantic Bird's-Eye View (BEV) segmentation. The models span:

3 training datasets — nuScenes, Woven Planet (Lyft), and a combined cross-dataset split
5 model architectures — from a camera-only baseline to multi-modal camera+LiDAR architectures
4 semantic segmentation tasks — vehicle, pedestrian, drivable area, and a joint vehicle+drivable area task

Each model weights file is located at {dataset}/{architecture}/{task}/model_best.pt and ships alongside its config.yaml.

Model Zoo

File Organization

{dataset}/
  {architecture}/
    {task}/
      model_best.pt     ← trained weights
      config.yaml       ← training configuration

Available Models

Dataset	Architecture	Vehicle	Human	Drivable Area	Vehicle + Drivable Area
lyft	`lss`	✓	✓	✓	✓
lyft	`lapt`	✓	✓	✓	✓
lyft	`lapt_fpn`	✓	✓	✓	✓
lyft	`lapt_pp`	✓	✓	✓	✓
lyft	`lapt_pp_fpn`	✓	✓	✓	✓
nuscenes	`lss`	✓	✓	✓	✓
nuscenes	`lapt`	✓	✓	✓	✓
nuscenes	`lapt_fpn`	✓	✓	✓	✓
nuscenes	`lapt_pp`	✓	✓	✓	✓
nuscenes	`lapt_pp_fpn`	✓	✓	✓	✓
nusc-lyft	`lss`	✓	✓	✓	✓
nusc-lyft	`lapt`	✓	✓	✓	✓
nusc-lyft	`lapt_fpn`	✓	✓	✓	✓
nusc-lyft	`lapt_pp`	✓	✓	✓	✓
nusc-lyft	`lapt_pp_fpn`	✓	✓	✓	✓

Architectures

All models output a semantic segmentation map in the Bird's-Eye View (BEV) frame over a 100 m × 100 m area centered on the ego-vehicle at a 0.5 m/cell resolution (200 × 200 grid).

LSS — Lift-Splat-Shoot (camera only)

Baseline camera-only architecture. Projects multi-camera features into a voxel representation using depth prediction, then encodes the voxels in BEV with a ResNet18-based decoder.

Backbone: EfficientNet-B0 with learned depth bins
Input: surround-view camera images (5–6 cameras)
Reference: Philion & Fidler, ECCV 2020

LAPT — Lidar-Aided Projection Transformer

Extends LSS by replacing the learned depth distribution with LiDAR-guided depth, improving geometric accuracy of the image-to-BEV projection.

Backbone: EfficientNet-B0 + LiDAR depth projection module
Input: surround-view cameras + LiDAR point cloud

LAPT-FPN — LAPT with Feature Pyramid Network

Adds a Feature Pyramid Network to the camera encoder for richer multi-scale image features before lifting into BEV.

Extras: dual-scale FPN (use_fpn: True)

LAPT-PP — LAPT with PointPillars

Fuses camera features (LAPT) with LiDAR features extracted by a PointPillars network. Features are combined via average-pooling fusion.

LiDAR encoder: Pillar Feature Network (PFN) with BN + ReLU
Fusion: avg_pool over camera + PointPillars feature maps

LAPT-PP-FPN — LAPT + PointPillars + FPN

Combines all modules: LAPT depth guidance, FPN multi-scale camera features, and PointPillars LiDAR features.

Usage

Installation

conda create -n beval python=3.10
conda activate beval
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
pip install efficientnet-pytorch==0.7.0 numba==0.57.1 nuscenes-devkit==1.1.9 \
            lyft-dataset-sdk==0.0.8 yacs==0.1.8 tensorboardx==2.6.2.2 \
            matplotlib==3.5.3 shapely==1.8.2

Clone the code repository:

git clone https://github.com/manueldiaz96/beval.git
cd beval

Set environment variables pointing to your dataset directories:

export NUSCENES=/path/to/nuscenes
export LYFT=/path/to/woven_planet

Note: The Woven Planet (Lyft) models require sub-sampled LiDAR point clouds.
Download them from manutheeng/subsampled-lyft and place them in $LYFT/subsampled_lidar/.

Running evaluation

Use the test script that matches the architecture:

# LSS (camera only)
python test_lift_splat.py \
    --cfg  path/to/model_zoo/nuscenes/lss/vehicle/config.yaml \
    --weights path/to/model_zoo/nuscenes/lss/vehicle/model_best.pt

# LAPT / LAPT-FPN
python test_LAPT.py \
    --cfg  path/to/model_zoo/nuscenes/lapt/vehicle/config.yaml \
    --weights path/to/model_zoo/nuscenes/lapt/vehicle/model_best.pt

# LAPT-PP / LAPT-PP-FPN
python test_LAPT_PP.py \
    --cfg  path/to/model_zoo/nuscenes/lapt_pp_fpn/vehicle/config.yaml \
    --weights path/to/model_zoo/nuscenes/lapt_pp_fpn/vehicle/model_best.pt

The script prints per-class IoU (%) for the validation split.

Loading weights directly in Python

import torch
from models.LAPTNet import compile_model  # or lift_splat / LAPTNet_PP

grid_conf    = {"xbound": [-50., 50., 0.5], "ybound": [-50., 50., 0.5]}
data_aug_conf = {"ncams": 6, "rand_flip": False, "pc_rot": (-20, 20),
                 "cams": ["CAM_FRONT_LEFT", "CAM_FRONT", "CAM_FRONT_RIGHT",
                          "CAM_BACK_RIGHT", "CAM_BACK", "CAM_BACK_LEFT"]}

model = compile_model(grid_conf, data_aug_conf, outC=1, use_fpn=False)
model.load_state_dict(torch.load("nuscenes/lapt/vehicle/model_best.pt"))
model.eval()

Training Data

Split	Dataset	Cameras	LiDAR
`lyft`	Woven Planet Perception Dataset (trainval)	5 surround cameras	✓ (sub-sampled)
`nuscenes`	nuScenes (trainval)	6 surround cameras	✓
`nusc-lyft`	nuScenes + Woven Planet (joint)	5–6 cameras	✓

BEV grid configuration (all models)

Parameter	Value
X / Y range	±50 m
Resolution	0.5 m/cell
Grid size	200 × 200

Results

Please refer to the paper for full quantitative results. The figures below are reproduced from the paper.

Cross-dataset evaluation — models trained on one dataset and evaluated on the other:

Cross-dataset evaluation results

Cross-dataset training — models trained on the combined nusc-lyft split:

Cross-dataset training results

Citation

If this model zoo is useful in your research, please cite:

@inproceedings{beval,
  title     = {BEVal: A Cross-dataset Evaluation Study of {BEV} Segmentation Models for Autonomous Driving},
  author    = {Diaz-Zapata, Manuel and Liu, Wenqian and Baruffa, Robin and Laugier, Christian},
  booktitle = {2024 18th International Conference on Control, Automation, Robotics and Vision (ICARCV)},
  pages     = {704--709},
  year      = {2024},
  organization = {IEEE}
}

License

This model zoo is released under the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) license.
Commercial use is not permitted.

Parts of the codebase are derived from Lift-Splat-Shoot (NVIDIA, MIT-licensed for the original code).

Sources

1 source

Hugging FacePrincipalVersion 40006d03b5e55bf4d428cd1c03419af2776dd91b

huggingface.co/manutheeng/beval_model_zoo/tree/40006d03b5e55bf4d428cd1c03419af2776dd91b

Vérifié 18 août

Artefacts du modèle

60 artefacts

lyft/lapt_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 b7fadbd7a80f…4473 · Hugging Face

Télécharger

lyft/lapt_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 0cbc51b2d3ad…deab · Hugging Face

Télécharger

lyft/lapt_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 19602eb9c98d…07f9 · Hugging Face

Télécharger

lyft/lapt_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 0acb5cb25541…bf9b · Hugging Face

Télécharger

lyft/lapt_pp_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 9f0bc6c008fe…4e52 · Hugging Face

Télécharger

lyft/lapt_pp_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 5d6e3fbec736…214f · Hugging Face

Télécharger

lyft/lapt_pp_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 bdc03610deec…6254 · Hugging Face

Télécharger

lyft/lapt_pp_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 04faf9089223…9b3b · Hugging Face

Télécharger

lyft/lapt_pp/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 fde7dfc6308f…0488 · Hugging Face

Télécharger

lyft/lapt_pp/human/model_best.pt

pt · 54,8 MB · SHA-256 5ea52ceac1a7…e9ab · Hugging Face

Télécharger

lyft/lapt_pp/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 05358af5edd0…9d64 · Hugging Face

Télécharger

lyft/lapt_pp/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 949f4b1af2c2…2d01 · Hugging Face

Télécharger

lyft/lapt/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 d70638938be8…a27d · Hugging Face

Télécharger

lyft/lapt/human/model_best.pt

pt · 54,8 MB · SHA-256 8ba30fdd5a27…fc63 · Hugging Face

Télécharger

lyft/lapt/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 adb79ad4d1e0…2907 · Hugging Face

Télécharger

lyft/lapt/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 6393c9a40b4c…ae9d · Hugging Face

Télécharger

lyft/lss/drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 27294eb1b277…6102 · Hugging Face

Télécharger

lyft/lss/human/model_best.pt

pt · 55,0 MB · SHA-256 a7ab48cab309…3af6 · Hugging Face

Télécharger

lyft/lss/vehicle_drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 c57a327940b4…a68d · Hugging Face

Télécharger

lyft/lss/vehicle/model_best.pt

pt · 55,0 MB · SHA-256 a1116a1a0057…9a89 · Hugging Face

Télécharger

nusc-lyft/lapt_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 0d4a172045a4…1c19 · Hugging Face

Télécharger

nusc-lyft/lapt_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 cbfdda0e0be1…0e82 · Hugging Face

Télécharger

nusc-lyft/lapt_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 f738e6d057ac…b0c1 · Hugging Face

Télécharger

nusc-lyft/lapt_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 8f71927241cb…46a9 · Hugging Face

Télécharger

nusc-lyft/lapt_pp_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 2338ecbf8c4c…ae56 · Hugging Face

Télécharger

nusc-lyft/lapt_pp_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 fe77b98052fa…830c · Hugging Face

Télécharger

nusc-lyft/lapt_pp_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 d543b1ab4435…ba62 · Hugging Face

Télécharger

nusc-lyft/lapt_pp_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 db9aae4a5513…7913 · Hugging Face

Télécharger

nusc-lyft/lapt_pp/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 3928dbecc57f…8ab3 · Hugging Face

Télécharger

nusc-lyft/lapt_pp/human/model_best.pt

pt · 54,8 MB · SHA-256 c627e905c15e…cdf3 · Hugging Face

Télécharger

nusc-lyft/lapt_pp/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 e61a35e5f7ac…bee7 · Hugging Face

Télécharger

nusc-lyft/lapt_pp/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 5b7b3902f746…7685 · Hugging Face

Télécharger

nusc-lyft/lapt/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 f7bb6f88aca1…fb35 · Hugging Face

Télécharger

nusc-lyft/lapt/human/model_best.pt

pt · 54,8 MB · SHA-256 a552ceacce01…533e · Hugging Face

Télécharger

nusc-lyft/lapt/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 5c368f217348…4831 · Hugging Face

Télécharger

nusc-lyft/lapt/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 f17eb48de454…a6a0 · Hugging Face

Télécharger

nusc-lyft/lss/drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 b416d0cf1c0a…f900 · Hugging Face

Télécharger

nusc-lyft/lss/human/model_best.pt

pt · 55,0 MB · SHA-256 070a8d707313…615c · Hugging Face

Télécharger

nusc-lyft/lss/vehicle_drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 9c53e8327a3a…2dd6 · Hugging Face

Télécharger

nusc-lyft/lss/vehicle/model_best.pt

pt · 55,0 MB · SHA-256 a6a46380e7f5…96b3 · Hugging Face

Télécharger

nuscenes/lapt_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 8e8a38fc61b4…b1ae · Hugging Face

Télécharger

nuscenes/lapt_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 4d4eb9d2cc4a…bdb1 · Hugging Face

Télécharger

nuscenes/lapt_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 6e8fd018cad8…9829 · Hugging Face

Télécharger

nuscenes/lapt_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 6e71140b1299…7852 · Hugging Face

Télécharger

nuscenes/lapt_pp_fpn/drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 9e28200bec67…45ab · Hugging Face

Télécharger

nuscenes/lapt_pp_fpn/human/model_best.pt

pt · 66,5 MB · SHA-256 37545bdb8f82…3f6d · Hugging Face

Télécharger

nuscenes/lapt_pp_fpn/vehicle_drivable_area/model_best.pt

pt · 66,5 MB · SHA-256 95c447502fab…fde5 · Hugging Face

Télécharger

nuscenes/lapt_pp_fpn/vehicle/model_best.pt

pt · 66,5 MB · SHA-256 95c26a138949…5dbb · Hugging Face

Télécharger

nuscenes/lapt_pp/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 d2f20cd71222…3022 · Hugging Face

Télécharger

nuscenes/lapt_pp/human/model_best.pt

pt · 54,8 MB · SHA-256 0306bdd1c982…9a0f · Hugging Face

Télécharger

nuscenes/lapt_pp/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 9c4ae7dbfa5d…f628 · Hugging Face

Télécharger

nuscenes/lapt_pp/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 7e826aff3f88…87b0 · Hugging Face

Télécharger

nuscenes/lapt/drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 766be851b753…28e5 · Hugging Face

Télécharger

nuscenes/lapt/human/model_best.pt

pt · 54,8 MB · SHA-256 ddefd4231714…348c · Hugging Face

Télécharger

nuscenes/lapt/vehicle_drivable_area/model_best.pt

pt · 54,8 MB · SHA-256 3354aefe7e0d…5da5 · Hugging Face

Télécharger

nuscenes/lapt/vehicle/model_best.pt

pt · 54,8 MB · SHA-256 eba6c554907e…5222 · Hugging Face

Télécharger

nuscenes/lss/drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 f4a6301fe85c…2e77 · Hugging Face

Télécharger

nuscenes/lss/human/model_best.pt

pt · 55,0 MB · SHA-256 66838426ee8a…0f37 · Hugging Face

Télécharger

nuscenes/lss/vehicle_drivable_area/model_best.pt

pt · 55,0 MB · SHA-256 d9752e2493b3…8c13 · Hugging Face

Télécharger

nuscenes/lss/vehicle/model_best.pt

pt · 55,0 MB · SHA-256 5f728571f5d6…7758 · Hugging Face

Télécharger

Extraits de sources

2 extraits

Readme18 aoûthuggingface.co/manutheeng/beval_model_zoo

--- license: cc-by-nc-4.0 task_categories: - image-segmentation tags: - autonomous-driving - bird's-eye-view - semantic-segmentation - bev-segmentation - lidar - multi-camera - pytorch - computer-vision - nuscenes - cross-dataset datasets: - nuscenes library_name: pytorch pipeline_tag: image-segmentation arxiv: 2408.16322 --- # BEVal Model Zoo — BEV Segmentation Models for Autonomous Driving This repository contains the pre-trained model weights from the paper: > **BEVal: A Cross-dataset Evaluation Study of BEV Segmentation Models for Autonomous Driving** > Manuel Diaz-Zapata, Wenqian Liu, Robin Baruffa, Christian Laugier > *2024 18th International Conference on Control, Automation, Robotics and Vision (ICARCV), pp. 704–709, IEEE.* > [![arXiv](https://img.shields.io/badge/arXiv-2408.16322-b31b1b.svg)](https://arxiv.org/abs/2408.16322) [![GitHub](https://img.shields.io/badge/GitHub-beval-181717?logo=github)](https://github.com/manueldiaz96/beval) --- ## Overview This collection contains **60 pre-trained models** for semantic Bird's-Eye View (BEV) segmentation. The models span: - **3 training datasets** — nuScenes, Woven Planet (Lyft), and a combined cross-dataset split - **5 model architectures** — from a camera-only baseline to multi-modal camera+LiDAR architectures - **4 semantic segmentation tasks** — vehicle, pedestrian, drivable area, and a joint vehicle+drivable area task Each model weights file is located at `{dataset}/{architecture}/{task}/model_best.pt` and ships alongside its `config.yaml`. --- ## Model Zoo ### File Organization ``` {dataset}/ {architecture}/ {task}/ model_best.pt ← trained weights config.yaml ← training configuration ``` ### Available Models | Dataset | Architecture | Vehicle | Human | Drivable Area | Vehicle + Drivable Area | |---------|-------------|:-------:|:-----:|:-------------:|:-----------------------:| | **lyft** | `lss` | ✓ | ✓ | ✓ | ✓ | | **lyft** | `lapt` | ✓ | ✓ | ✓ | ✓ | | **lyft** | `lapt_fpn` | ✓ | ✓ | ✓ | ✓ | | **lyft** | `lapt_pp` | ✓ | ✓ | ✓ | ✓ | | **lyft** | `lapt_pp_fpn` | ✓ | ✓ | ✓ | ✓ | | **nuscenes** | `lss` | ✓ | ✓ | ✓ | ✓ | | **nuscenes** | `lapt` | ✓ | ✓ | ✓ | ✓ | | **nuscenes** | `lapt_fpn` | ✓ | ✓ | ✓ | ✓ | | **nuscenes** | `lapt_pp` | ✓ | ✓ | ✓ | ✓ | | **nuscenes** | `lapt_pp_fpn` | ✓ | ✓ | ✓ | ✓ | | **nusc-lyft** | `lss` | ✓ | ✓ | ✓ | ✓ | | **nusc-lyft** | `lapt` | ✓...

Note18 aoûthuggingface.co/manutheeng/beval_model_zoo

Source context: 0 downloads · 0 likes · Pipeline image-segmentation · Library pytorch · Repo manutheeng/beval_model_zoo