-
Notifications
You must be signed in to change notification settings - Fork 974
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
add paddle3d and bevfusion model training #2749
base: develop
Are you sure you want to change the base?
Conversation
## 4. 数据准备和校验 | ||
### 4.1 数据准备 | ||
|
||
本教程采用 `nuScenes`数据集 作为示例数据集,请在[官网](https://www.nuscenes.org/nuscenes)进行下载,将数据集目录准备如下: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里最好把示例数据集上传到BOS上吧?
|—— v1.0-trainval | ||
``` | ||
|
||
在`./paddlex/repo_manager/repos/Paddle3D`的目录下创建软链接 `data/nuscenes`,指向到上面的数据集目录: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
其实不太希望让用户在这个目录下创建内容,PaddleX对大多数用户来说,希望这个repos是无感的。这块看是否挪出去呢?
mkdir data | ||
ln -s /path/to/nuscenes_dataset_root ./data | ||
mv ./data/nuscenes_dataset_root ./data/nuscenes | ||
``` |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
关于用户自己的数据如何制作和标注,需要补充文档说明下
|
||
BEVFusion模型训练需要加载camera分支模型预训练分支和lidar分支模型预训练权重: | ||
```bash | ||
wget https://paddle3d.bj.bcebos.com/models/bevfusion/camera/model.pdparams |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这一块需要内置到代码中,在config中配置,尽量不让用户手动执行这些操作
model: BEVFusion | ||
mode: check_dataset | ||
dataset_dir: "/paddle/dataset/paddlex/3d/nuscenes" | ||
device: gpu:1,2,3,4 #0,1,2,3 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里为什么是1,2,3,4呢?
|
||
|
||
Export: | ||
weight_path: /paddle/model/paddlex/3d/fusion.pdparams |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里应该写动态图权重的http链接
|
||
Evaluate: | ||
batch_size: 1 | ||
weight_path: /paddle/model/paddlex/3d/fusion.pdparams |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里写对应的output的动态图路径
@@ -0,0 +1,96 @@ | |||
# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
2025
|
||
|
||
class BEVFusionTrainer(BaseTrainer): | ||
"""Object Detection Model Trainer""" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
注释不对
@@ -0,0 +1,17 @@ | |||
# Copyright (c) 2023 PaddlePaddle Authors. All Rights Reserved. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
2025,下同
add paddle3d and bevfusion model training. Original pr #2572