README

offline_camera_detection

offline_camera_detection is Apollo's camera offline detection tool.

How to work

1. First, build the perception module

./apollo.sh build perception

2. Then run the tool

bash modules/perception/tools/offline_camera_detection/offline_camera_detection.sh

3. Finally, view the results in /apollo/data

test.jpg  test.txt

Input

The input image data is in the images directory. You need to put the test pictures into this directory, and add the picture name in images\image_test_list.txt.

The default test image is images\test.jpg, and images\image_test_list.txt contains the name test in it.

Output

The output data is in /apollo/data, Include text files and image files.

The default test result is test.jpg test.txt.

Model config

The model config file is in conf\camera_detection_pipeline.pb.txt, which you can change the model.

The default detection model is SMOKE_OBSTACLE_DETECTION.

Parameters

offline_camera_detection supports the following parameters

• height, Input image height
• width, Input image width
• image_ext, Input image format