An implementation of the yolov3 custom convolutional layer. Parameters ---------- name : string The name of the tensor to b

Creates the typical residual block architecture. Residual blocks are useful for training very deep convolutional neural networks because the

Returns a random number between a and b.

Takes the box_data and maps it into the y_true numpy array for a particular grid size. The yolov3 model has three grid siz

Calculates and returns the loss associated with a particular layer scale. Parameters ---------- name : string

Takes the output of a convolutional layer and transforms it into fully scaled bounding box coordinates. These predicted boxes can then be pas

Reads the anchors from a file and returns them in a list. Parameters ---------- anchors_path : string Path that points to wh

Reads the class names from a file and returns them in a list. Parameters ---------- classes_path : string Path that points t

Takes the annotion lines, reads them, and from their information constructs the necessary numpy arrays that store the data to train the yolo

Takes the inputs tensor and route tensor and then concatenates them along the last axis (the channel axis). Parameters ----------

Takes the outputs of the previous convolutional layer and upsamples them by a factor of two using the 'nearest neighbor' method. Paramet

The fully defined architecture of the version three yolo model. Parameters ---------- inputs : tensor An array of shape:

Takes the iamge and box data and applies random transformations if the 'random' parameter is set to true. Otherwise, the image and box data w

Converts coordinates in the form of center_x, center_y, width, height to min_x, min_y, max_x, max_y. The coordinate values are already scale

A wrapper function for creating the full y_true and y_true_box_data numpy arrays used for training the yolov3 model. Pa

Draws identified boxes along with class probabilities on the original image and then saves the image with the output file name. Paramete

Calculates the intersection over union (IOU) of two bounding boxes, which is the ratio of the area where the two boxes overlap to th

Calculates the IOU (Intersection over Union) between the predicted boxes (prct_yolo_outputs) and the true boxes (y_true_boxes

Applies non-max suppression to predicted boxes. Parameters ---------- predictions_with_boxes : ndarray An array of shape:

Takes the raw data from the text file and splits it up into a training set and a validation set based on the train/val split hyperparameter.

Takes any image and transforms it into the format needed for object detection with yolov3. Parameters ---------- image_path : string

A wrapper function that returns the loss associated with a forward pass of the yolo_v3 model. The main purpose of this function is to extract