Omni-MOT Dataset Recorder

This is the repository of recording Omni-MOT dataset. Its functions include:

• Camera Operations: Move forward backward, pitch up down, roll left right, yaw left right
• Save camera parameters
• Calculate the ground truth data for multiple object tracking (i.e. 3D bounding boxes, 2D bounding boxes, Visibility)
• Save Videos and Ground Truth

Requirement

• CARLA 0.9.4, refer to [CARLA Docs]

• >= Ubuntu 16.0

• clone this repository into your local disk

  git clone https://github.com/shijieS/OMOTDRecorder.git

• Install the required packages

  cd <this repository>
  pip install -r requirement.txt

• make sure all the required packages are installed and your python environment is activated.

Usage

• launch CARLA Simulator

cd <CARLA Project>
./CarlaUE4.sh

Setup Camera

• set camera by running

  python start_recording.py --config_name=test.json --auto_save=False --flag_show_windows=True --recording_num_scale=0

  or

  ./select_viewpoint.sh

  use the short cut key to move camera and fix its viewpoint.

Recording

• modify the configure files as you want, in our example, we use test.json which is generated by "Setup Camera"

  {
      "cameras": {
          "Easy_Camera_Cross1": {
              "fov": 90.0,
              "height": 1080,
              "max_record_frame": 10,
              "pitch": -42.999786376953125,
              "roll": 1.0506457329029217e-05,
              "width": 1920,
              "x": -79.3550796508789,
              "y": -0.8948922753334045,
              "yaw": 42.000396728515625,
              "z": 32.788326263427734
          },
          "Hard_Town3_Test_00": {
              "fov": 90.0,
              "height": 1080,
              "max_record_frame": 10000,
              "pitch": 0.0,
              "roll": 0.0,
              "width": 1920,
              "x": -78.0,
              "y": 8.300000190734863,
              "yaw": 0.0,
              "z": 39.0
          }
      },
      "host": "127.0.0.1",
      "mode": "parallel", 
      "port": 2000,
      "save_root": "/home/ssj/Data/github/AwesomeMOTDataset/Dataset/Temp",
      "vehicle_num": [
          30,
          45,
          50
      ],
      "weathers": [
          "Clear",
          "Cloudy",
          "Rain"
      ]
  }

• Run a script

  python start_recording.py --config_name=test.json --auto_save=True --flag_show_windows=True --recording_num_scale=0.5

  or

  ./start_recording.sh

Short Cut Key

Key	Action
Esc	Quit
Insert	Insert a new camera
Enter	Save all camera parameters
Tab	Switch the Cameras
1	Start Image Processing
2	Show 3D Bounding Boxes in the Image
3	Show Rectangle in the Image
4	Show Vehicle Labels
w	Move Forward
s	Move Backward
a	Move Left
d	Move Right
q	Move Up
e	Move Down
i	Pitch Up
k	Pitch Down
j	Roll Left
l	Roll Right
u	Yaw Left
p	Yaw Right
Home	Update the Camera Level

Citation

@inproceedings{ShiJie20,
  author = {Shijie Sun, Naveed Aktar, XiangYu Song, Huansheng Song, Ajmal Mian, Mubarak Shah},
  title = {Simultaneous Detection and Tracking with Motion Modelling for Multiple Object Tracking},
  booktitle = {Proceedings of the European conference on computer vision (ECCV)}},
  year = {2020}

@inproceedings{Dosovitskiy17,
  title = {{CARLA}: {An} Open Urban Driving Simulator},
  author = {Alexey Dosovitskiy and German Ros and Felipe Codevilla and Antonio Lopez and Vladlen Koltun},
  booktitle = {Proceedings of the 1st Annual Conference on Robot Learning},
  pages = {1--16},
  year = {2017}
}

Acknowledge

This work is based on the CARLA . It is also inspired by SSD and DAN.

License

CARLA specific code is distributed under MIT License.

CARLA specific assets are distributed under CC-BY License.

Note that UE4 itself follows its own license terms.