From 7ef3e810155d8cf1344e9da84662063669ca1d2b Mon Sep 17 00:00:00 2001 From: Zhijian Liu Date: Fri, 3 Jun 2022 21:22:22 -0400 Subject: [PATCH] Update news and results --- README.md | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/README.md b/README.md index 5cca03e1..a97ab18e 100644 --- a/README.md +++ b/README.md @@ -8,8 +8,10 @@ **If you are interested in getting updates, please sign up [here](https://docs.google.com/forms/d/e/1FAIpQLSfkmfsX45HstL5rUQlS7xJthhS3Z_Pm2NOVstlXUqgaK4DEfQ/viewform) to get notified!** +- **(2022/6/3)** BEVFusion ranks first on [nuScenes](https://nuscenes.org/object-detection?externalData=all&mapData=all&modalities=Any) among all solutions. - **(2022/6/3)** We released the first version of BEVFusion (with pre-trained checkpoints and evaluation). -- **(2022/5/26)** BEVFusion is released on arXiv. +- **(2022/5/26)** BEVFusion is released on [arXiv](https://arxiv.org/abs/2205.13542). +- **(2022/5/2)** BEVFusion ranks first on [nuScenes](https://nuscenes.org/object-detection?externalData=all&mapData=all&modalities=Any) among all solutions that do not use test-time augmentation and model ensemble. ## Abstract @@ -17,13 +19,14 @@ Multi-sensor fusion is essential for an accurate and reliable autonomous driving ## Results -### 3D Object Detection on nuScenes test +### 3D Object Detection (on nuScenes test) | Model | Modality | mAP | NDS | | :-------: | :------: | :--: | :--: | -| [BEVFusion](configs/nuscenes/det/transfusion/secfpn/camera+lidar/swint_v0p075/convfuser.yaml) | C+L | 70.23 | 72.88 | +| BEVFusion-e | C+L | 74.99 | 76.09 | +| BEVFusion | C+L | 70.23 | 72.88 | -### 3D Object Detection on nuScenes validation +### 3D Object Detection (on nuScenes validation) | Model | Modality | mAP | NDS | Checkpoint | | :------------------: | :------: | :--: | :--: | :---------: | @@ -33,7 +36,7 @@ Multi-sensor fusion is essential for an accurate and reliable autonomous driving *Note*: The camera-only object detection baseline is a variant of BEVDet-Tiny with a much heavier view transformer and other differences in hyperparameters. Thanks to our [efficient BEV pooling](mmdet3d/ops/bev_pool) operator, this model runs fast and has higher mAP than BEVDet-Tiny under the same input resolution. Please refer to [BEVDet repo](https://github.com/HuangJunjie2017/BEVDet) for the original BEVDet-Tiny implementation. The LiDAR-only baseline is TransFusion-L. -### BEV Map Segmentation on nuScenes validation +### BEV Map Segmentation (on nuScenes validation) | Model | Modality | mIoU | Checkpoint | | :------------------: | :------: | :--: | :---------: |