holding bounding boxes for ground truth and detections

holding bounding boxes for ground truth and detections

holding bounding boxes for ground truth and detections

holding data needed for precision-recall and precision-aos

holding data needed for precision-recall and precision-aos

holding data needed for precision-recall and precision-aos

Contains modified VGG model definition to extract features from Bird's eye view input using pyramid features.

Static class to return preconfigured dataset objects

Frame Calibration Holder 3x4 p0-p3 Camera P matrix. Contains extrinsic and intrinsic parameters. 3x

Modified VGG model definition to extract features from RGB image input using pyramid features.

Object Label Class 1 type Describes the type of object: 'Car', 'Van', 'Truck', 'Pedestrian', 'Person_sitting', 'C

Stereo Calibration Holder 1 baseline distance between the two camera centers. 1 f focal length. 3x3 k intr

Abstract base class for anchor generators.

Subsamples minibatches to a desired balance of positives and negatives.

This is used in vgg unit tests.

Box collection.

evaluated object classes

evaluated object classes

evaluated object classes

easy, moderate and hard evaluation level

easy, moderate and hard evaluation level

easy, moderate and hard evaluation level

Abstract base class for detection models.

Abstract base class for loss functions.

evaluation metrics: image, ground or 3D

evaluation metrics: image, ground or 3D

evaluation metrics: image, ground or 3D

Abstract base class for subsampling minibatches.

Enum class for sort order. Attributes: ascend: ascend order. descend: descend order.

L2 localization loss function with anchorwise output support. Loss[b,a] = .5 * ||weights[b,a] * (prediction[b,a,:] - target[b,a,:])||^2

Sigmoid cross entropy classification loss function.

Smooth L1 localization loss function. The smooth L1_loss is defined elementwise as .5 x^2 if |x|<1 and |x|-.5 otherwise, where x is the differ

Softmax cross-entropy loss function.

Softmax cross-entropy loss function.

holding detection data

holding detection data

holding detection data

holding ground truth data

holding ground truth data

holding ground truth data