Models

The Live Sense models provide an interface for interacting with the underlying machine learning (ML) models used for object detection.

Info: Object Detection/Recognition

Describes the combination of object classification, identifying if objects of interest exist in an image, and localization, identifying where the object is located in the image. For example, a Live Sense model that identifies a vehicle in the image and provides a bounding box of where the vehicle was identified.

The core function of every model is to process an RGB image and return a list of detections found within the image.

Each Live Sense model has a set of object classes that it is able to detect. The developer must determine which objects and models are relevant to their use case.

Model types and detection classes

The following table describes the models and their available classes:

Category Model Name Label Recommended Confidence
Road Basics RB_day pedestrian 47
rider 47
bicycle 45
car 46
motorcycle 47
bus 50
truck 45
traffic-light 47
RB_night pedestrian 45
car 40
motorcycle 40
truck 40
Road Alerts BrakeLight car-brake-light-on 80
Road Hazards cone_barrier cone 60
strip-stand 60
rectangular-barrier 75
cylindrical-barrier 60
delineator-barrier 60
pothole pothole 80
speed_bump_signage speed-bump-sign 55
Road Signs RoadSigns DS_SpeedLimit[5-130] 55
DS_AdvisorySpeedLimit[10-85] 55
DS_Stop 55
DS_DoNotEnter 55
DS_NoAccess 55
DS_StartConstruction 55
DS_PedestrianCrossing 55
DS_SpeedBumps 55
DS_StopHereOnRed 55
DS_NoLeftTurn 55
DS_NoRightOrLeftTurn 55
DS_NoUTurn 55
DS_NoUTurnOrLeftTurn 55
DS_ParkingRestriction 55

Note

RB_night detects supported features in low light conditions.

 

Note

For the Road Signs model, only the label prefix is given above.
See the Road Signs section below for the full label list with sample images

Models available in beta mode

The models below are available for beta testing:

Category Model Name Label Recommended Confidence
Road Alerts TrafficLightCVModel traffic-light-green 75
traffic-light-red 75
traffic-light-yellow 75
Road Hazards bridge_tunnel bridge 75
tunnel 75
signage roadworks-going-on 60
road-closed 55
speed_bump_object cross-walk 55
height_restriction_signs height-restriction-sign-Xft-Yin 60
Lane Detection lane_detector lane N/A
RTG real_time_guidance stop-sign 60
traffic-light 60

Confidence configuration

The Live Sense SDK allows you to configure the confidence values of the models as well as the individual classes within each model.

Update confidence of a model

The following function updates the confidence of all classes available in the specified model to the provided value.

func updateModelConfidence(modelKey : String, modelConfidence : Float)

The modelKey can be used from the table described in the Model Types and Detection Classes section.

Update confidence of a class

The following function updates the confidence of the specified class to the provided value. This supersedes the confidence configured for the parent model.

func updateClassConfidence(classLabel : ClassLabelName, classConfidence : Float)

The classLabel can be used from the table described in the Model Types and Detection Classes section.

Get confidence of a model or a class

func getConfidenceThreshold(modelKey : String, classLabel:ClassLabelName?) -> Float

The modelKey and classLabel can be used from the table described in the Model Types and Detection Classes section.

Warning :: Limitations

Using a lower confidence threshold will result in more detections, but an overall lower accuracy. For this reason, applications should keep their confidence threshold at or above the recommended value to maintain a reasonably accurate output.

Detections work better in Landscape mode.

Note

  • In addition, these methods can be called during a running session.
  • For the following labels, class-wise confidence is not supported:
    • brake-light-on
    • DS_SpeedLimit[5-130]_White_Circle_00
    • DS_SpeedLimit[5-130]_White_Rectangle_01

Model execution options

Live Sense SDK supports multiple instances for executing the Live Sense models. Exact support depends upon the device and model being used.

Note

If a configuration cannot be used as-is, the SDK will fall back to the next best supported configuration.

Object recognition

A Live Sense LSDRecognition describes the following properties of a detected object:

  • class - Which object was detected. For more information, see Model Types and Detection Classes.
  • location - Where the object was found in the image frame.
  • confidence score - A number between 0 and 1 that indicates confidence that the object was correctly detected.

Model runtime support for underlying architecture

CPU GPU/Metal (A6 Chip) ANE(A12 Chip)
Road Basics YES YES YES
Road Alerts YES YES YES
Road Signs YES YES YES
Road Hazards YES YES YES

Distance and position estimation

Note

This feature is in Beta, the returned distance and relative position values may be inaccurate.

All of the object detection models listed above can provide an estimated distance and position relative to the camera's point of view for each detected object.

The output values for the distance and relativePosition properties of a detected object are as follows:

  • Height: Distance of the detection from the ground.
  • Lateral: Distance of the detection relative to the center of the camera's view. This value may be positive or negative with a negative value signifying that the detection is towards the left side of the camera's view and vise-versa.
  • Depth: Distance that the device needs to travel forward in a straight line so that the actual distance is equal to the lateral distance.
  • Distance: Actual distance of the detection from the camera's point of view.

All distance values are provided in meters and are used in the following features of Live Sense SDK

  • Time-to-collision alerts
  • Relative velocity estimation

Road signs

Group Label Sample Image
SpeedLimit DS_SpeedLimitXX_White_Circle_00
DS_SpeedLimitXX_White_Circle_01
DS_SpeedLimitXX_White_Rectangle_01
AdvisorySpeedLimit DS_AdvisorySpeedLimitXX_Yellow_Rectangle_00
DS_AdvisorySpeedLimitXX_Yellow_Rectangle_01
DS_AdvisorySpeedLimitXX_Yellow_Rectangle_03
Stop DS_Stop_Red_Octogon_00
DoNotEnter DS_DoNotEnter_Red_Circle_00
DS_DoNotEnter_White_Rectangle_00
DS_DoNotEnter_White_Rectangle_01
NoAccess DS_NoAccess_White_Circle_00
StartConstruction DS_StartConstruction_White_Triangle_00
Yield DS_Yield_White_Triangle_00
PedstrianCrossing DS_PedestrianCrossing_Blue_Rectangle_00
DS_PedestrianCrossing_White_Triangle_00
DS_PedestrianCrossing_Yellow_Circle_00
DS_PedestrianCrossing_Yellow_Diamond_01
DS_PedestrianCrossing_Yellow_Triangle_00
SpeedBumps DS_SpeedBumps_Blue_Rectangle_00
DS_SpeedBumps_White_Triangle_00
DS_SpeedBumps_Yellow_Diamond_01
StopHereOnRed DS_StopHereOnRed_White_Rectangle_00
NoLeftTurn DS_NoLeftTurn_White_Circle_00
DS_NoLeftTurn_White_Rectangle_02
NoRightOrLeftTurn DS_NoRightOrLeftTurn_White_Rectangle_00
DS_NoRightOrLeftTurn_White_Circle_00
NoUTurn DS_NoUTurn_White_Circle_00
NoUTurnOrLeftTurn DS_NoUTurnOrLeftTurn_White_Rectangle_00
ParkingRestriction DS_ParkingRestriction_Blue_Circle_00
DS_ParkingRestriction_White_Circle_01
DS_ParkingRestriction_White_Rectangle_03

results matching ""

    No results matching ""