offlinemot.detection.YoloDetector.better_detection

YoloDetector.better_detection(imgfile, additional_objs=[])

Detect the classes within the image at the proposed areas based on the previous detections. This will make yolo network take higher resolution input. Additionally the full scale detections are included.

Parameters

• imgfile (str or numpy array) – The input image that could be a string filename or a numpy array

• additional_objs (list) – The list of current tracked and moving objects boxes. These will form the cropped parts that the model will perfrom detections on.

Returns

A list of detection positions and classes and the shape of the original image. The list has the following shape [top-left point, bottom-right point,probabilty, class id]

Return type

list,(int,int)