sliding_window.py

################################################################################

# functionality: functions for multi-scale sliding window (exhaustive) search

# This version: (c) 2018 Toby Breckon, Dept. Computer Science, Durham University, UK
# License: MIT License

# Origin acknowledgements: forked from https://github.com/siphomateke/PyBOW

################################################################################

import numpy as np
import cv2

################################################################################

# re-size an image with respect to its aspect ratio if needed.
# used in the multi-scale image pyramid approach

def resize_img(img, width=-1, height=-1):
    if height == -1 and width == -1:
        raise TypeError("Invalid arguments. Width or height must be provided.")
    h = img.shape[0]
    w = img.shape[1]
    if height == -1:
        aspect_ratio = float(w) / h
        new_height = int(width / aspect_ratio)
        return cv2.resize(img, (width, new_height))
    elif width == -1:
        aspect_ratio = h / float(w)
        new_width = int(height / aspect_ratio)
        return cv2.resize(img, (new_width, height))

################################################################################

# a very basic approach to produce an image at multi-scales (i.e. variant
# re-sized resolutions)

def pyramid(img, scale=1.5, min_size=(30, 30)):
    # yield the original image
    yield img

    # keep looping over the pyramid
    while True:
        # compute the new dimensions of the image and resize it
        w = int(img.shape[1] / scale)
        img = resize_img(img, width=w)

        # if the resized image does not meet the supplied minimum
        # size, then stop constructing the pyramid
        if img.shape[0] < min_size[1] or img.shape[1] < min_size[0]:
            break

        # yield the next image in the pyramid
        yield img

################################################################################

# generate a set of sliding window locations across the image

def sliding_window(image, window_size, step_size=8):
    # slide a window across the image
    for y in range(0, image.shape[0], step_size):
        for x in range(0, image.shape[1], step_size):
            # yield the current window
            window = image[y:y + window_size[1], x:x + window_size[0]]
            if not (window.shape[0] != window_size[1] or window.shape[1] != window_size[0]):
                yield (x, y, window)

################################################################################

# perform basic non-maximal suppression of overlapping object detections

def non_max_suppression_fast(boxes, overlapThresh):
    # if there are no boxes, return an empty list
    if len(boxes) == 0:
        return []

    # if the bounding boxes integers, convert them to floats --
    # this is important since we'll be doing a bunch of divisions
    if boxes.dtype.kind == "i":
        boxes = boxes.astype("float")

    # initialize the list of picked indexes
    pick = []

    # grab the coordinates of the bounding boxes
    x1 = boxes[:, 0]
    y1 = boxes[:, 1]
    x2 = boxes[:, 2]
    y2 = boxes[:, 3]

    # compute the area of the bounding boxes and sort the bounding
    # boxes by the bottom-right y-coordinate of the bounding box
    area = (x2 - x1 + 1) * (y2 - y1 + 1)
    idxs = np.argsort(y2)

    # keep looping while some indexes still remain in the indexes
    # list
    while len(idxs) > 0:
        # grab the last index in the indexes list and add the
        # index value to the list of picked indexes
        last = len(idxs) - 1
        i = idxs[last]
        pick.append(i)

        # find the largest (x, y) coordinates for the start of
        # the bounding box and the smallest (x, y) coordinates
        # for the end of the bounding box
        xx1 = np.maximum(x1[i], x1[idxs[:last]])
        yy1 = np.maximum(y1[i], y1[idxs[:last]])
        xx2 = np.minimum(x2[i], x2[idxs[:last]])
        yy2 = np.minimum(y2[i], y2[idxs[:last]])

        # compute the width and height of the bounding box
        w = np.maximum(0, xx2 - xx1 + 1)
        h = np.maximum(0, yy2 - yy1 + 1)

        # compute the ratio of overlap
        overlap = (w * h) / area[idxs[:last]]

        # delete all indexes from the index list that have a significant overlap
        idxs = np.delete(idxs, np.concatenate(([last],
                                               np.where(overlap > overlapThresh)[0])))

    # return only the bounding boxes that were picked using the
    # integer data type
    return boxes[pick].astype("int")

################################################################################