jaccard.py

# ----------------------------------------------------------------------------
# A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation
#-----------------------------------------------------------------------------
# Copyright (c) 2016 Federico Perazzi
# Licensed under the BSD License [see LICENSE for details]
# Written by Federico Perazzi
# ----------------------------------------------------------------------------

""" Compute Jaccard Index. """

import numpy as np
import glob
import os
from PIL import Image
from options import OPTION as opt

def db_eval_iou(annotation,segmentation):

	""" Compute region similarity as the Jaccard Index.
	Arguments:
		annotation   (ndarray): binary annotation   map.
		segmentation (ndarray): binary segmentation map.
	Return:
		jaccard (float): region similarity
 """

	annotation   = annotation.astype(np.bool)
	segmentation = segmentation.astype(np.bool)

	if np.isclose(np.sum(annotation),0) and np.isclose(np.sum(segmentation),0):
		return 1
	else:
		return np.sum((annotation & segmentation)) / \
				np.sum((annotation | segmentation),dtype=np.float32)


def eval_jaccard():
    root_dir = 'data/'
    pred_dir = root_dir + opt.output_dir +'/MSD/'
    gt_dir = root_dir + 'MSD/Task10_mask/'
    
    filelist = os.listdir(pred_dir) 
    jaccards = np.zeros(len(filelist))
    for i in range(0, len(filelist)):

        item = os.path.join(pred_dir,filelist[i])
        preds = os.listdir(item)
        _mask = []
        _gt = []
        for ii in range(len(preds)):
            filename = os.path.join(item, preds[ii])
            mask = np.array(Image.open(filename)).astype(np.float32)
            gt = np.array(Image.open(os.path.join(gt_dir, filelist[i],preds[ii].split('-')[1]))).astype(np.float32)
            _mask.append(mask)
            _gt.append(gt)
        jaccards[i] = db_eval_iou(np.stack(_gt,0), np.stack(_mask,0))

    J = jaccards.mean()
    print("jaccard:",J)
    return J
if __name__ == '__main__':
    eval_jaccard()