chore: add project template and update document (#15)

Signed-off-by: scepter914 <[email protected]>

Author: scepter914

README.md

@@ -20,45 +20,40 @@ This repository is ML library for Autoware.
 ### Tools
 
 - 3D
-  - [x] Train code
-  - [ ] Evaluation code
   - [ ] Make mp4 files for visualization
   - [ ] Make mcap files for visualization
 - 2D
-  - [ ] Train code
-  - [ ] Evaluation code
   - [ ] Make mp4 files for visualization
   - [ ] Make mcap files for visualization
 
 ### 3D Detection model
 
-- [BEVFusion (Camera-LiDAR fusion)](projects/BEVFusion)
-- TransFusion (LiDAR only)
+- [BEVFusion-CL (BEVFusion Camera-LiDAR fusion)](projects/BEVFusion)
+- TransFusion-L (TransFusion LiDAR only)
 - CenterPoint (LiDAR only)
 - PointPainted-CenterPoint (Camera-LiDAR fusion)
 
-|                | Onnx deploy | T4dataset | NuScenes | aimotive |
-| -------------- | :---------: | :-------: | :------: | :------: |
-| BEVFusion-CL   |             |           |    ✅     |          |
-| TransFusion-L  |             |           |          |          |
-| CenterPoint    |             |           |          |          |
-| PP-CenterPoint |             |           |          |          |
+|                | T4dataset | NuScenes | aimotive |
+| -------------- | :-------: | :------: | :------: |
+| BEVFusion-CL   |           |          |          |
+| TransFusion-L  |           |          |          |
+| CenterPoint    |           |          |          |
+| PP-CenterPoint |           |          |          |
 
 ### 2D Detection
 
 - YOLOX-opt
 - TwinTransformer
 
-|                 | Onnx deploy | T4dataset | NuImages | COCO  |
-| --------------- | :---------: | :-------: | :------: | :---: |
-| YOLOX-opt       |             |           |          |       |
-| TwinTransformer |             |           |          |       |
+|                 | T4dataset | NuImages | COCO  |
+| --------------- | :-------: | :------: | :---: |
+| YOLOX-opt       |           |          |       |
+| TwinTransformer |           |          |       |
 
 ### 2D classification
 
 - EfficientNet
 
-|              | Onnx deploy | T4dataset | NuImages | COCO  |
-| ------------ | :---------: | :-------: | :------: | :---: |
-| EfficientNet |             |           |          |       |
-
+|              | T4dataset | NuImages | COCO  |
+| ------------ | :-------: | :------: | :---: |
+| EfficientNet |           |          |       |

docs/release_note.md

@@ -0,0 +1,10 @@
+## For next release
+
+- Add BEVFusion model (#1, #2, #12)
+- Update docs (#3, #4, #5, #6, #7, #8, #10, #11, #13)
+- Add dataset converter (#9)
+- Add train code for 2D (#14)
+
+## v0.1.0
+
+TBD

projects/BEVFusion/README.md

@@ -1,3 +1,12 @@
+## Supported feature
+
+- [x] Train LiDAR-only model
+- [ ] Train Camera-LiDAR fusion model
+- [x] Train with single GPU
+- [ ] Train with multi GPU
+- [ ] Make .onnx file and deploy to Autoware
+- [ ] Unit test
+
 ## Usage
 ### Setup
 

projects/template/README.md

@@ -0,0 +1,30 @@
+## Supported feature
+
+- [ ] Train LiDAR-only model
+- [ ] Train Camera-LiDAR fusion model
+- [ ] Train with single GPU
+- [ ] Train with multi GPU
+- [ ] Make .onnx file and deploy to Autoware
+- [ ] Unit test
+
+## Usage
+### Setup
+
+TBD
+
+### Train
+
+TBD
+
+### Deploy
+
+TBD
+
+## Results and models
+### NuScenes
+
+TBD
+
+### T4 dataset
+
+TBD

projects/template/nuscenes/template_1xb1-cyclic-20e_nus-3d.py

@@ -0,0 +1,5 @@
+_base_ = [
+    '../../../../configs/fcos3d/fcos3d_r101-caffe-dcn_fpn_head-gn_8xb2-1x_nus-mono3d.py'  # noqa
+]
+custom_imports = dict(imports=['projects.example_project.dummy'])
+_base_.model.backbone.type = 'DummyResNet'