zl3466 commited on
Commit
d903b08
1 Parent(s): 0f41e58

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -1
README.md CHANGED
@@ -43,6 +43,7 @@ This tutorial explains how the NuScenes structure works in our dataset, includin
43
  - [Scene](#scene)
44
  - [Sample](#sample)
45
  - [Sample Data](#sample-data)
 
46
  - [Camera](#camera-data)
47
  - [LiDAR](#lidar-data)
48
  - [IMU](#imu-data)
@@ -78,6 +79,7 @@ nusc = NuScenes(version='v1.0', dataroot=f'/MARS_multiagent', verbose=True)
78
 
79
  <br/>
80
 
 
81
  ## Scene
82
  To see all scenes in one set (one location of the Multitraversal set, or the whole Multiagent set):
83
  ```
@@ -152,6 +154,7 @@ Output:
152
  <br/>
153
 
154
  ## Sample Data
 
155
  Our sensor names are different from NuScenes' sensor names. It is important that you use the correct name when querying sensor data. Our sensor names are:
156
  ```
157
  ['CAM_FRONT_CENTER',
@@ -166,6 +169,9 @@ Our sensor names are different from NuScenes' sensor names. It is important that
166
 
167
  ---
168
  ### Camera Data
 
 
 
169
  ```
170
  sensor = 'CAM_FRONT_CENTER'
171
  sample_data_token = my_sample['data'][sensor]
@@ -198,8 +204,12 @@ Output:
198
  - `prev`: previous data token for this sensor
199
  - `next`: next data token for this sensor
200
 
201
- All image data are already undistorted. You may now load the image in any ways you'd like. Here's an example using cv2:
 
 
202
  ```
 
 
203
  data_path, boxes, camera_intrinsic = nusc.get_sample_data(sample_data_token)
204
  img = cv2.imread(data_path)
205
  cv2.imshow('fc_img', img)
@@ -228,6 +238,7 @@ The 5-dimensional data array is in `pcd.points`. Below is an example of visualiz
228
 
229
 
230
  ```
 
231
  from nuscenes.utils.data_classes import LidarPointCloud
232
 
233
  sensor = 'LIDAR_FRONT_CENTER'
 
43
  - [Scene](#scene)
44
  - [Sample](#sample)
45
  - [Sample Data](#sample-data)
46
+ - [Sensor Names](#sensor-names)
47
  - [Camera](#camera-data)
48
  - [LiDAR](#lidar-data)
49
  - [IMU](#imu-data)
 
79
 
80
  <br/>
81
 
82
+
83
  ## Scene
84
  To see all scenes in one set (one location of the Multitraversal set, or the whole Multiagent set):
85
  ```
 
154
  <br/>
155
 
156
  ## Sample Data
157
+ ### Sensor Names
158
  Our sensor names are different from NuScenes' sensor names. It is important that you use the correct name when querying sensor data. Our sensor names are:
159
  ```
160
  ['CAM_FRONT_CENTER',
 
169
 
170
  ---
171
  ### Camera Data
172
+ All image data are already undistorted.
173
+
174
+ To load a piece data, we start with querying its `sample_data` dictionary object from the metadata:
175
  ```
176
  sensor = 'CAM_FRONT_CENTER'
177
  sample_data_token = my_sample['data'][sensor]
 
204
  - `prev`: previous data token for this sensor
205
  - `next`: next data token for this sensor
206
 
207
+ After getting the `sample_data` dictionary, Use NuScenes devkit's `get_sample_data()` function to retrieve the data's absolute path.
208
+
209
+ Then you may now load the image in any ways you'd like. Here's an example using `cv2`:
210
  ```
211
+ import cv2
212
+
213
  data_path, boxes, camera_intrinsic = nusc.get_sample_data(sample_data_token)
214
  img = cv2.imread(data_path)
215
  cv2.imshow('fc_img', img)
 
238
 
239
 
240
  ```
241
+ import open3d as o3d
242
  from nuscenes.utils.data_classes import LidarPointCloud
243
 
244
  sensor = 'LIDAR_FRONT_CENTER'