pyCarDisplay package

Subpackages

• pyCarDisplay.detection package
  • Subpackages
    • pyCarDisplay.detection.midas package
      • Submodules
      • pyCarDisplay.detection.midas.base_model module
      • pyCarDisplay.detection.midas.blocks module
      • pyCarDisplay.detection.midas.midas_net module
      • pyCarDisplay.detection.midas.midas_net_custom module
      • pyCarDisplay.detection.midas.transforms module
      • Module contents
  • Submodules
  • pyCarDisplay.detection.depth_detection_api module
  • pyCarDisplay.detection.object_detection_api module
  • Module contents
• pyCarDisplay.sensors package
  • Submodules
  • pyCarDisplay.sensors.image_processing_api module
  • pyCarDisplay.sensors.imu_api module
  • pyCarDisplay.sensors.kalman_filter_api module
  • Module contents
• pyCarDisplay.utils package
  • Submodules
  • pyCarDisplay.utils.display_api module
  • pyCarDisplay.utils.kitti_data_loader_api module
  • Module contents

Submodules

pyCarDisplay.car module

All of the Car operations are controlled from this file.

class pyCarDisplay.car.Car(car_images_path: str, imu_sensor_path='', lidar_sensor_path='', object_detection_model_path='', depth_detection_model_path='', depth_detection_model_type='large', optimize=True, dpi=100, alpha=0.6, img_resize_size=(300, 300), pixel_sizes=[1242, 375], norm_mean=[0.485, 0.456, 0.406], norm_std=[0.229, 0.224, 0.225], R_covariance=0.1, add_noise=True, IMU_name=None, gui_speed=1, theme='DarkGrey1', image_extension='png', random_state=42, verbose=False, device='cpu', track_n_frames=10, plot_column='af')

Bases: object

Initilize the Car class.

car_images_pathstr

Path to the Kitti dataset image frames.

imu_sensor_pathstr, optional

Path to the Kitti dataset IMU sensor data. The default is "". If not provided, simulation will not work on the IMU data.

lidar_sensor_pathstr, optional

Path to the Kitti dataset Lidar sensor data. The default is "". If not provided, simulation will not work on the Lidar data.

object_detection_model_pathstr, optional

Path to the trained PyTorch object detection model. The default is "". If not provided, object detection is not performed.

depth_detection_model_pathstr, optional

Path to the trained PyTorch depth detection model. The default is "".

depth_detection_model_typestr, optional

If 'large', model uses a larger trained model but it will be slower. The default is "large".

optimizebool, optional

If True, images are pre-processed for better depth detection. The default is True.

dpiint, optional

Dots per inch. The default is 100.

alphadouble, optional

The alpha variable. The default is 0.6.

img_resize_sizetuple, optional

Size of the image when performing object detection. The default is (300, 300).

pixel_sizeslist, optional

The sizes of the pixels. The default is [1242, 375].

norm_meanlist, optional

Object detection hyper-parameter. The default is [0.485, 0.456, 0.406].

norm_stdlist, optional

Object detection hyper-parameter. The default is [0.229, 0.224, 0.225].

R_covariancefloat, optional

Error covariance for IMU data. The default is 0.1.

add_noisebool, optional

If True, random uniform noise added to the IMU data. The default is True.

gui_speedint, optional

Time between each frame in GUI. The default is 1.

themestr, optional

GUI color theme. The default is "DarkGrey1".

image_extensionstr, optional

Kitti dataset image extension. The default is "png". Used when reading the images from the path.

random_stateint, optional

random seed. The default is 42.

verbosebool, optional

Flag to print information. The default is False.

devicestr, optional

If 'gpu' passed, ML modules will use the GPU. The default is "cpu". CUDA device must be available.

track_n_framesint, optional

How many frames to track in the Kalman plot. The default is 10.

plot_columnstr, optional

Which IMU column to plot in the Kalman plot. The default is "af".

None.

run(verbose=None)

Iterates the taken images and one by one performs object detection and depth detection on the objects renders the imgaes taken on a display with the depth heatmaps as subimages.

verboseBool

Flag to print more information at runtime.

None. Displays GUI.

set_frame(frame: int)

Manually sets the frame.

frameint

Target frame.

None.

pyCarDisplay.pyCarDisplay module

pyCarDisplay is a autonomous vehicle simulator. It is designed to simulate driving car based on the datasets provided in Kitti (http://www.cvlibs.net/datasets/kitti/). pyCarDisplay can show the image frames in a Kitti dataset on a GUI, perform object detection given a trained model, perform depth detection given a trained model, simulate IMU sensor given IMU data, and use Kalman filter.

pyCarDisplay class is used as an API wrapper to the rest of the modules, i.e. the simulation is invoked from this file.

class pyCarDisplay.pyCarDisplay.CarDisplay(**parameters)

Bases: object

The main API class of the pyCarDisplay.

Initilize the CarDisplay class. This class passes the below parameters to the Car class.

The initlization also sets up the environment variables.

car_images_pathstr

Path to the Kitti dataset image frames.

imu_sensor_pathstr, optional

Path to the Kitti dataset IMU sensor data. The default is "". If not provided, simulation will not work on the IMU data.

lidar_sensor_pathstr, optional

Path to the Kitti dataset Lidar sensor data. The default is "". If not provided, simulation will not work on the Lidar data.

object_detection_model_pathstr, optional

Path to the trained PyTorch object detection model. The default is "". If not provided, object detection is not performed.

depth_detection_model_pathstr, optional

Path to the trained PyTorch depth detection model. The default is "".

depth_detection_model_typestr, optional

If 'large', model uses a larger trained model but it will be slower. The default is "large".

optimizebool, optional

If True, images are pre-processed for better depth detection. The default is True.

dpiint, optional

Dots per inch. The default is 100.

alphadouble, optional

The alpha variable. The default is 0.6.

img_resize_sizetuple, optional

Size of the image when performing object detection. The default is (300, 300).

pixel_sizeslist, optional

The sizes of the pixels. The default is [1242, 375].

norm_meanlist, optional

Object detection hyper-parameter. The default is [0.485, 0.456, 0.406].

norm_stdlist, optional

Object detection hyper-parameter. The default is [0.229, 0.224, 0.225].

R_covariancefloat, optional

Error covariance for IMU data. The default is 0.1.

add_noisebool, optional

If True, random uniform noise added to the IMU data. The default is True.

gui_speedint, optional

Time between each frame in GUI. The default is 1.

themestr, optional

GUI color theme. The default is "DarkGrey1".

image_extensionstr, optional

Kitti dataset image extension. The default is "png". Used when reading the images from the path.

random_stateint, optional

random seed. The default is 42.

verbosebool, optional

Flag to print information. The default is False.

devicestr, optional

If 'gpu' passed, ML modules will use the GPU. The default is "cpu". CUDA device must be available.

track_n_framesint, optional

How many frames to track in the Kalman plot. The default is 10.

plot_columnstr, optional

Which IMU column to plot in the Kalman plot. The default is "af".

None.

get_params()

Can be used to get the object variables.

self

The object variables in dict format.

set_params(verbose=True, **parameters)

Can be used to change the Car settings.

verbosebool, optional

Used the show warning. The default is True.

**parametersdict

all the object variables accepted by the car class.

None.

start(**parameters)

Iterates the taken images and one by one performs object detection and depth detection on the objects renders the imgaes taken on a display with the depth heatmaps as subimages.

verboseBool

Flag to print more information at runtime.

None. Displays GUI.

pyCarDisplay.version module

pyCarDisplay version.

Module contents