getting_started.md 12.4 KB
Newer Older
twang's avatar
twang committed
1
# Prerequisites
zhangwenwei's avatar
zhangwenwei committed
2

3
- Linux or macOS (Windows is in experimental support)
twang's avatar
twang committed
4
5
6
7
- Python 3.6+
- PyTorch 1.3+
- CUDA 9.2+ (If you build PyTorch from source, CUDA 9.0 is also compatible)
- GCC 5+
xiliu8006's avatar
xiliu8006 committed
8
9
10
- [MMCV](https://mmcv.readthedocs.io/en/latest/#installation)


11
12
13
14
The required versions of MMCV, MMDetection and MMSegmentation for different versions of MMDetection3D are as below. Please install the correct version of MMCV, MMDetection and MMSegmentation to avoid installation issues.

| MMDetection3D version | MMDetection version | MMSegmentation version |    MMCV version     |
|:-------------------:|:-------------------:|:-------------------:|:-------------------:|
15
| master              | mmdet>=2.19.0, <=3.0.0| mmseg>=0.20.0, <=1.0.0 | mmcv-full>=1.4.8, <=1.5.0|
16
17
18
| v1.0.0rc0           | mmdet>=2.19.0, <=3.0.0| mmseg>=0.20.0, <=1.0.0 | mmcv-full>=1.3.17, <=1.5.0|
| 0.18.1              | mmdet>=2.19.0, <=3.0.0| mmseg>=0.20.0, <=1.0.0 | mmcv-full>=1.3.17, <=1.5.0|
| 0.18.0              | mmdet>=2.19.0, <=3.0.0| mmseg>=0.20.0, <=1.0.0 | mmcv-full>=1.3.17, <=1.5.0|
Wenhao Wu's avatar
Wenhao Wu committed
19
20
21
22
23
24
25
26
27
28
29
30
31
32
| 0.17.3              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.17.2              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.17.1              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.17.0              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.16.0              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.15.0              | mmdet>=2.14.0, <=3.0.0| mmseg>=0.14.1, <=1.0.0 | mmcv-full>=1.3.8, <=1.4.0|
| 0.14.0              | mmdet>=2.10.0, <=2.11.0| mmseg==0.14.0 | mmcv-full>=1.3.1, <=1.4.0|
| 0.13.0              | mmdet>=2.10.0, <=2.11.0| Not required  | mmcv-full>=1.2.4, <=1.4.0|
| 0.12.0              | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.2.4, <=1.4.0|
| 0.11.0              | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.2.4, <=1.3.0|
| 0.10.0              | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.2.4, <=1.3.0|
| 0.9.0               | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.2.4, <=1.3.0|
| 0.8.0               | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.1.5, <=1.3.0|
| 0.7.0               | mmdet>=2.5.0, <=2.11.0 | Not required  | mmcv-full>=1.1.5, <=1.3.0|
Tai-Wang's avatar
Tai-Wang committed
33
| 0.6.0               | mmdet>=2.4.0, <=2.11.0 | Not required  | mmcv-full>=1.1.3, <=1.2.0|
34
| 0.5.0               | 2.3.0                  | Not required  | mmcv-full==1.0.5|
zhangwenwei's avatar
Doc  
zhangwenwei committed
35

twang's avatar
twang committed
36
# Installation
zhangwenwei's avatar
Doc  
zhangwenwei committed
37

twang's avatar
twang committed
38
## Install MMDetection3D
zhangwenwei's avatar
Doc  
zhangwenwei committed
39

40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
### Quick installation instructions script

Assuming that you already have CUDA 11.0 installed, here is a full script for quick installation of MMDetection3D with conda.
Otherwise, you should refer to the step-by-step installation instructions in the next section.

```shell
conda create -n open-mmlab python=3.7 pytorch=1.9 cudatoolkit=11.0 torchvision -c pytorch -y
conda activate open-mmlab
pip3 install openmim
mim install mmcv-full
mim install mmdet
mim install mmsegmentation
git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
pip3 install -e .
```

### Step-by-step installation instructions

59
**a. Create a conda virtual environment and activate it.**
zhangwenwei's avatar
zhangwenwei committed
60

twang's avatar
twang committed
61
62
63
```shell
conda create -n open-mmlab python=3.7 -y
conda activate open-mmlab
zhangwenwei's avatar
Doc  
zhangwenwei committed
64
65
```

66
**b. Install PyTorch and torchvision following the [official instructions](https://pytorch.org/).**
Wenwei Zhang's avatar
Wenwei Zhang committed
67

twang's avatar
twang committed
68
69
```shell
conda install pytorch torchvision -c pytorch
Wenwei Zhang's avatar
Wenwei Zhang committed
70
71
```

twang's avatar
twang committed
72
73
Note: Make sure that your compilation CUDA version and runtime CUDA version match.
You can check the supported CUDA version for precompiled packages on the [PyTorch website](https://pytorch.org/).
Wenwei Zhang's avatar
Wenwei Zhang committed
74

75
`E.g. 1` If you have CUDA 10.1 installed under `/usr/local/cuda` and would like to install
twang's avatar
twang committed
76
PyTorch 1.5, you need to install the prebuilt PyTorch with CUDA 10.1.
Wenwei Zhang's avatar
Wenwei Zhang committed
77

twang's avatar
twang committed
78
```python
79
conda install pytorch==1.5.0 cudatoolkit=10.1 torchvision==0.6.0 -c pytorch
Wenwei Zhang's avatar
Wenwei Zhang committed
80
81
```

twang's avatar
twang committed
82
83
`E.g. 2` If you have CUDA 9.2 installed under `/usr/local/cuda` and would like to install
PyTorch 1.3.1., you need to install the prebuilt PyTorch with CUDA 9.2.
zhangwenwei's avatar
zhangwenwei committed
84

twang's avatar
twang committed
85
86
```python
conda install pytorch=1.3.1 cudatoolkit=9.2 torchvision=0.4.2 -c pytorch
wangtai's avatar
wangtai committed
87
88
```

89
If you build PyTorch from source instead of installing the prebuilt package,
twang's avatar
twang committed
90
you can use more CUDA versions such as 9.0.
91

92
**c. Install [MMCV](https://mmcv.readthedocs.io/en/latest/).**
xiliu8006's avatar
xiliu8006 committed
93
*mmcv-full* is necessary since MMDetection3D relies on MMDetection, CUDA ops in *mmcv-full* are required.
zhangwenwei's avatar
Doc  
zhangwenwei committed
94

95
`e.g.` The pre-build *mmcv-full* could be installed by running: (available versions could be found [here](https://mmcv.readthedocs.io/en/latest/#install-with-pip))
zhangwenwei's avatar
zhangwenwei committed
96

Ziyi Wu's avatar
Ziyi Wu committed
97
```shell
xiliu8006's avatar
xiliu8006 committed
98
99
100
101
102
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/{cu_version}/{torch_version}/index.html
```

Please replace `{cu_version}` and `{torch_version}` in the url to your desired one. For example, to install the latest `mmcv-full` with `CUDA 11` and `PyTorch 1.7.0`, use the following command:

twang's avatar
twang committed
103
```shell
xiliu8006's avatar
xiliu8006 committed
104
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu110/torch1.7.0/index.html
twang's avatar
twang committed
105
```
zhangwenwei's avatar
zhangwenwei committed
106

107
108
109
110
111
112
113
mmcv-full is only compiled on PyTorch 1.x.0 because the compatibility usually holds between 1.x.0 and 1.x.1. If your PyTorch version is 1.x.1, you can install mmcv-full compiled with PyTorch 1.x.0 and it usually works well.

```shell
# We can ignore the micro version of PyTorch
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu110/torch1.7/index.html
```

xiliu8006's avatar
xiliu8006 committed
114
See [here](https://github.com/open-mmlab/mmcv#install-with-pip) for different versions of MMCV compatible to different PyTorch and CUDA versions.
twang's avatar
twang committed
115
Optionally, you could also build the full version from source:
zhangwenwei's avatar
zhangwenwei committed
116

twang's avatar
twang committed
117
```shell
xiliu8006's avatar
xiliu8006 committed
118
119
120
121
122
123
124
125
126
127
git clone https://github.com/open-mmlab/mmcv.git
cd mmcv
MMCV_WITH_OPS=1 pip install -e .  # package mmcv-full will be installed after this step
cd ..
```

Or directly run

```shell
pip install mmcv-full
twang's avatar
twang committed
128
```
zhangwenwei's avatar
zhangwenwei committed
129

130
**d. Install [MMDetection](https://github.com/open-mmlab/mmdetection).**
zhangwenwei's avatar
zhangwenwei committed
131

twang's avatar
twang committed
132
```shell
133
pip install mmdet
twang's avatar
twang committed
134
```
zhangwenwei's avatar
zhangwenwei committed
135

twang's avatar
twang committed
136
Optionally, you could also build MMDetection from source in case you want to modify the code:
zhangwenwei's avatar
zhangwenwei committed
137
138

```shell
twang's avatar
twang committed
139
140
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
141
git checkout v2.19.0  # switch to v2.19.0 branch
twang's avatar
twang committed
142
143
pip install -r requirements/build.txt
pip install -v -e .  # or "python setup.py develop"
zhangwenwei's avatar
zhangwenwei committed
144
145
```

146
147
148
**e. Install [MMSegmentation](https://github.com/open-mmlab/mmsegmentation).**

```shell
149
pip install mmsegmentation
150
151
152
153
154
155
156
```

Optionally, you could also build MMSegmentation from source in case you want to modify the code:

```shell
git clone https://github.com/open-mmlab/mmsegmentation.git
cd mmsegmentation
157
git checkout v0.20.0  # switch to v0.20.0 branch
158
159
160
161
pip install -e .  # or "python setup.py develop"
```

**f. Clone the MMDetection3D repository.**
zhangwenwei's avatar
Doc  
zhangwenwei committed
162

twang's avatar
twang committed
163
164
165
166
```shell
git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
```
zhangwenwei's avatar
zhangwenwei committed
167

168
**g.Install build requirements and then install MMDetection3D.**
zhangwenwei's avatar
zhangwenwei committed
169

twang's avatar
twang committed
170
171
172
```shell
pip install -v -e .  # or "python setup.py develop"
```
zhangwenwei's avatar
zhangwenwei committed
173

twang's avatar
twang committed
174
Note:
zhangwenwei's avatar
Doc  
zhangwenwei committed
175

twang's avatar
twang committed
176
177
1. The git commit id will be written to the version number with step d, e.g. 0.6.0+2e7045c. The version will also be saved in trained models.
It is recommended that you run step d each time you pull some updates from github. If C++/CUDA codes are modified, then this step is compulsory.
zhangwenwei's avatar
Doc  
zhangwenwei committed
178

twang's avatar
twang committed
179
    > Important: Be sure to remove the `./build` folder if you reinstall mmdet with a different CUDA/PyTorch version.
zhangwenwei's avatar
zhangwenwei committed
180

twang's avatar
twang committed
181
182
183
184
185
    ```shell
    pip uninstall mmdet3d
    rm -rf ./build
    find . -name "*.so" | xargs rm
    ```
zhangwenwei's avatar
zhangwenwei committed
186

187
2. Following the above instructions, MMDetection3D is installed on `dev` mode, any local modifications made to the code will take effect without the need to reinstall it (unless you submit some commits and want to update the version number).
zhangwenwei's avatar
zhangwenwei committed
188

twang's avatar
twang committed
189
190
3. If you would like to use `opencv-python-headless` instead of `opencv-python`,
you can install it before installing MMCV.
zhangwenwei's avatar
zhangwenwei committed
191

twang's avatar
twang committed
192
4. Some dependencies are optional. Simply running `pip install -v -e .` will only install the minimum runtime requirements. To use optional dependencies like `albumentations` and `imagecorruptions` either install them manually with `pip install -r requirements/optional.txt` or specify desired extras when calling `pip` (e.g. `pip install -v -e .[optional]`). Valid keys for the extras field are: `all`, `tests`, `build`, and `optional`.
zhangwenwei's avatar
zhangwenwei committed
193

twang's avatar
twang committed
194
5. The code can not be built for CPU only environment (where CUDA isn't available) for now.
zhangwenwei's avatar
zhangwenwei committed
195

twang's avatar
twang committed
196
## Another option: Docker Image
Wenwei Zhang's avatar
Wenwei Zhang committed
197

twang's avatar
twang committed
198
We provide a [Dockerfile](https://github.com/open-mmlab/mmdetection3d/blob/master/docker/Dockerfile) to build an image.
Wenwei Zhang's avatar
Wenwei Zhang committed
199

twang's avatar
twang committed
200
201
```shell
# build an image with PyTorch 1.6, CUDA 10.1
202
docker build -t mmdetection3d -f docker/Dockerfile .
twang's avatar
twang committed
203
```
Wenwei Zhang's avatar
Wenwei Zhang committed
204

twang's avatar
twang committed
205
Run it with
Wenwei Zhang's avatar
Wenwei Zhang committed
206

twang's avatar
twang committed
207
208
209
```shell
docker run --gpus all --shm-size=8g -it -v {DATA_DIR}:/mmdetection3d/data mmdetection3d
```
Wenwei Zhang's avatar
Wenwei Zhang committed
210

twang's avatar
twang committed
211
## A from-scratch setup script
Wenwei Zhang's avatar
Wenwei Zhang committed
212

213
Here is a full script for setting up MMdetection3D with conda.
Wenwei Zhang's avatar
Wenwei Zhang committed
214

twang's avatar
twang committed
215
216
217
```shell
conda create -n open-mmlab python=3.7 -y
conda activate open-mmlab
Wenwei Zhang's avatar
Wenwei Zhang committed
218

219
# install latest PyTorch prebuilt with the default prebuilt CUDA version (usually the latest)
twang's avatar
twang committed
220
conda install -c pytorch pytorch torchvision -y
Wenwei Zhang's avatar
Wenwei Zhang committed
221

twang's avatar
twang committed
222
223
# install mmcv
pip install mmcv-full
liyinhao's avatar
liyinhao committed
224

twang's avatar
twang committed
225
226
# install mmdetection
pip install git+https://github.com/open-mmlab/mmdetection.git
liyinhao's avatar
liyinhao committed
227

228
229
230
# install mmsegmentation
pip install git+https://github.com/open-mmlab/mmsegmentation.git

twang's avatar
twang committed
231
232
233
234
# install mmdetection3d
git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
pip install -v -e .
zhangwenwei's avatar
zhangwenwei committed
235
```
liyinhao's avatar
liyinhao committed
236

twang's avatar
twang committed
237
238
239
## Using multiple MMDetection3D versions

The train and test scripts already modify the `PYTHONPATH` to ensure the script use the MMDetection3D in the current directory.
liyinhao's avatar
liyinhao committed
240

twang's avatar
twang committed
241
242
243
244
To use the default MMDetection3D installed in the environment rather than that you are working with, you can remove the following line in those scripts

```shell
PYTHONPATH="$(dirname $0)/..":$PYTHONPATH
liyinhao's avatar
liyinhao committed
245
246
```

twang's avatar
twang committed
247
# Verification
liyinhao's avatar
liyinhao committed
248

249
## Verify with point cloud demo
zhangwenwei's avatar
Doc  
zhangwenwei committed
250

251
We provide several demo scripts to test a single sample. Pre-trained models can be downloaded from [model zoo](model_zoo.md). To test a single-modality 3D detection on point cloud scenes:
zhangwenwei's avatar
Doc  
zhangwenwei committed
252
253

```shell
wuyuefeng's avatar
Demo  
wuyuefeng committed
254
python demo/pcd_demo.py ${PCD_FILE} ${CONFIG_FILE} ${CHECKPOINT_FILE} [--device ${GPU_ID}] [--score-thr ${SCORE_THR}] [--out-dir ${OUT_DIR}]
zhangwenwei's avatar
Doc  
zhangwenwei committed
255
256
257
258
259
```

Examples:

```shell
260
python demo/pcd_demo.py demo/data/kitti/kitti_000008.bin configs/second/hv_second_secfpn_6x8_80e_kitti-3d-car.py checkpoints/hv_second_secfpn_6x8_80e_kitti-3d-car_20200620_230238-393f000c.pth
zhangwenwei's avatar
zhangwenwei committed
261
```
262

yinchimaoliang's avatar
yinchimaoliang committed
263
If you want to input a `ply` file, you can use the following function and convert it to `bin` format. Then you can use the converted `bin` file to generate demo.
264
Note that you need to install `pandas` and `plyfile` before using this script. This function can also be used for data preprocessing for training ```ply data```.
265

yinchimaoliang's avatar
yinchimaoliang committed
266
267
268
269
270
```python
import numpy as np
import pandas as pd
from plyfile import PlyData

271
def convert_ply(input_path, output_path):
yinchimaoliang's avatar
yinchimaoliang committed
272
273
274
275
276
277
278
279
280
281
    plydata = PlyData.read(input_path)  # read file
    data = plydata.elements[0].data  # read data
    data_pd = pd.DataFrame(data)  # convert to DataFrame
    data_np = np.zeros(data_pd.shape, dtype=np.float)  # initialize array to store data
    property_names = data[0].dtype.names  # read names of properties
    for i, name in enumerate(
            property_names):  # read data by property
        data_np[:, i] = data_pd[name]
    data_np.astype(np.float32).tofile(output_path)
```
282

yinchimaoliang's avatar
yinchimaoliang committed
283
Examples:
zhangwenwei's avatar
zhangwenwei committed
284

yinchimaoliang's avatar
yinchimaoliang committed
285
286
287
```python
convert_ply('./test.ply', './test.bin')
```
zhangwenwei's avatar
zhangwenwei committed
288

289
If you have point clouds in other format (`off`, `obj`, etc.), you can use `trimesh` to convert them into `ply`.
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304

```python
import trimesh

def to_ply(input_path, output_path, original_type):
    mesh = trimesh.load(input_path, file_type=original_type)  # read file
    mesh.export(output_path, file_type='ply')  # convert to ply
```

Examples:

```python
to_ply('./test.obj', './test.ply', 'obj')
```

305
More demos about single/multi-modality and indoor/outdoor 3D detection can be found in [demo](demo.md).
306

twang's avatar
twang committed
307
## High-level APIs for testing point clouds
zhangwenwei's avatar
zhangwenwei committed
308

twang's avatar
twang committed
309
### Synchronous interface
Ziyi Wu's avatar
Ziyi Wu committed
310

liyinhao's avatar
liyinhao committed
311
Here is an example of building the model and test given point clouds.
zhangwenwei's avatar
zhangwenwei committed
312
313

```python
314
from mmdet3d.apis import init_model, inference_detector
zhangwenwei's avatar
zhangwenwei committed
315

liyinhao's avatar
liyinhao committed
316
317
config_file = 'configs/votenet/votenet_8x8_scannet-3d-18class.py'
checkpoint_file = 'checkpoints/votenet_8x8_scannet-3d-18class_20200620_230238-2cea9c3a.pth'
zhangwenwei's avatar
zhangwenwei committed
318
319

# build the model from a config file and a checkpoint file
320
model = init_model(config_file, checkpoint_file, device='cuda:0')
zhangwenwei's avatar
zhangwenwei committed
321
322

# test a single image and show the results
liyinhao's avatar
liyinhao committed
323
324
325
326
point_cloud = 'test.bin'
result, data = inference_detector(model, point_cloud)
# visualize the results and save the results in 'results' folder
model.show_results(data, result, out_dir='results')
zhangwenwei's avatar
zhangwenwei committed
327
```