generate_anchors.py

import random
import argparse
import numpy as np

from detection_dataloader import parse_voc_annotation
import json

def IOU(ann, centroids):
    w, h = ann
    similarities = []

    for centroid in centroids:
        c_w, c_h = centroid

        if c_w >= w and c_h >= h:
            similarity = w*h/(c_w*c_h)
        elif c_w >= w and c_h <= h:
            similarity = w*c_h/(w*h + (c_w-w)*c_h)
        elif c_w <= w and c_h >= h:
            similarity = c_w*h/(w*h + c_w*(c_h-h))
        else: #means both w,h are bigger than c_w and c_h respectively
            similarity = (c_w*c_h)/(w*h)
        similarities.append(similarity) # will become (k,) shape

    return np.array(similarities)

def avg_IOU(anns, centroids):
    n,d = anns.shape
    sum = 0.

    for i in range(anns.shape[0]):
        sum+= max(IOU(anns[i], centroids))

    return sum/n

def print_anchors(centroids):
    out_string = ''

    anchors = centroids.copy()

    widths = anchors[:, 0]
    sorted_indices = np.argsort(widths)

    r = "anchors: ["
    for i in sorted_indices:
        out_string += str(int(anchors[i,0]*416)) + ',' + str(int(anchors[i,1]*416)) + ', '
            
    print(out_string[:-2])

def run_kmeans(ann_dims, anchor_num):
    ann_num = ann_dims.shape[0]
    iterations = 0
    prev_assignments = np.ones(ann_num)*(-1)
    iteration = 0
    old_distances = np.zeros((ann_num, anchor_num))

    indices = [random.randrange(ann_dims.shape[0]) for i in range(anchor_num)]
    centroids = ann_dims[indices]
    anchor_dim = ann_dims.shape[1]

    while True:
        distances = []
        iteration += 1
        for i in range(ann_num):
            d = 1 - IOU(ann_dims[i], centroids)
            distances.append(d)
        distances = np.array(distances) # distances.shape = (ann_num, anchor_num)

        print("iteration {}: dists = {}".format(iteration, np.sum(np.abs(old_distances-distances))))

        #assign samples to centroids
        assignments = np.argmin(distances,axis=1)

        if (assignments == prev_assignments).all() :
            return centroids

        #calculate new centroids
        centroid_sums=np.zeros((anchor_num, anchor_dim), np.float)
        for i in range(ann_num):
            centroid_sums[assignments[i]]+=ann_dims[i]
        for j in range(anchor_num):
            centroids[j] = centroid_sums[j]/(np.sum(assignments==j) + 1e-6)

        prev_assignments = assignments.copy()
        old_distances = distances.copy()

def _main_():
    num_anchors = 9
    
    root = 'F:\\Learning\\tensorflow\\detect\\Dataset\\'
    '''
    train_imgs, train_labels = parse_voc_annotation(
        root+'VOC2012\\Annotations\\',
        root+'VOC2012\\JPEGImages\\',
        'data.pkl',
        ['person','head','hand','foot','aeroplane','tvmonitor','train','boat','dog','chair',
         'bird','bicycle','bottle','sheep','diningtable','horse','motorbike','sofa','cow',
         'car','cat','bus','pottedplant']
    )
    '''
    '''
    train_imgs, train_labels = parse_voc_annotation(
        root+'Fish\\Annotations\\',
        root+'Fish\\JPEGImages\\',
        'data.pkl',
        ['heidiao','niyu','lvqimamiantun','hualu','heijun','dalongliuxian','tiaoshiban']
    )
    '''
    train_imgs, train_labels = parse_voc_annotation(
        root+'Fish\\Annotations\\',
        root+'Fish\\JPEGImages\\',
        'data.pkl',
        ['aeroplane','bicycle','bird','boat','bottle','bus','car','cat','chair','cow','diningtable','dog','horse','motorbike',
        'person','pottedplant','sheep','sofa','train','tvmonitor']
    )
    # run k_mean to find the anchors
    annotation_dims = []
    for image in train_imgs:
        print(image['filename'])
        for obj in image['object']:
            relative_w = (float(obj['xmax']) - float(obj['xmin']))/image['width']
            relatice_h = (float(obj["ymax"]) - float(obj['ymin']))/image['height']
            annotation_dims.append(tuple(map(float, (relative_w,relatice_h))))
    
    annotation_dims = np.array(annotation_dims)
    centroids = run_kmeans(annotation_dims, num_anchors)

    # write anchors to file
    print('\naverage IOU for', num_anchors, 'anchors:', '%0.2f' % avg_IOU(annotation_dims, centroids))
    print_anchors(centroids)

if __name__ == '__main__':
    _main_()
    #raccon 117,142, 151,233, 188,341, 245,378, 248,223, 288,302, 349,379, 374,274, 383,387
    #voc 24,35, 48,87, 71,192, 120,293, 126,100, 173,186, 218,330, 330,193, 361,362
    #    24,34, 46,84, 68,185, 116,286, 122,97, 171,180, 214,327, 326,193, 359,359
    #voc 18,27, 28,75, 49,132, 55,43, 65,227, 84,86, 108,162, 109,288, 162,329, 174,103, 190,212, 245,348
    #    321,150, 343,256, 372,379
    #fish 25,31, 36,52, 42,96, 59,33, 62,64, 65,134, 100,229, 107,82, 111,134, 159,180, 162,308, 223,131,
    #    255,198, 272,331, 412,412
    #fish 33,36, 53,69, 71,144, 106,85, 137,151, 145,269, 240,173, 257,326, 412,412