mAPコンセプトリファレンス https://blog.csdn.net/u013066730/article/details/82707559
import os
import numpy as np
from os import mkdir
import xml.etree.ElementTree as ET
from easydict import EasyDict as edict
cfgs = edict()
cfgs.EVALUATE_DIR = r"E:\data\yolov3\test\evaluate"
cfgs.VERSION = r"cls_dir"
cfgs.USE_07_METRIC = False
NAME_LABEL_MAP = {"person": 0, "car": 1}
def write_voc_results_file(all_boxes, test_imgid_list, det_save_dir):
for cls, cls_id in NAME_LABEL_MAP.items():
if cls == 'back_ground':
continue
print("Writing {} VOC resutls file".format(cls))
if not os.path.exists(det_save_dir):
mkdir(det_save_dir)
det_save_path = os.path.join(det_save_dir, "det_" + cls + ".txt")
with open(det_save_path, 'wt') as f:
for index, img_name in enumerate(test_imgid_list):
this_img_detections = np.array(all_boxes[index])
if this_img_detections.shape[0] == 0:
continue
this_cls_detections = this_img_detections[this_img_detections[:, 0] == cls_id]
if this_cls_detections.shape[0] == 0:
continue # this cls has none detections in this img
for a_det in this_cls_detections:
f.write('{:s} {:.3f} {:.1f} {:.1f} {:.1f} {:.1f}\n'.
format(img_name, a_det[1],
a_det[2], a_det[3],
a_det[4], a_det[5])) # that is [img_name, score, xmin, ymin, xmax, ymax]
def parse_rec(filename):
""" Parse a PASCAL VOC xml file """
tree = ET.parse(filename)
objects = []
for obj in tree.findall('object'):
obj_struct = {}
obj_struct['name'] = obj.find('name').text #####################这里可能需要调整
obj_struct['pose'] = obj.find('pose').text
obj_struct['truncated'] = int(obj.find('truncated').text)
obj_struct['difficult'] = int(obj.find('difficult').text)
bbox = obj.find('bndbox')
obj_struct['bbox'] = [int(bbox.find('xmin').text),
int(bbox.find('ymin').text),
int(bbox.find('xmax').text),
int(bbox.find('ymax').text)]
objects.append(obj_struct)
return objects
def voc_ap(rec, prec, use_07_metric=False):
if use_07_metric:
# 11 point metric
ap = 0.
for t in np.arange(0., 1.1, 0.1):
if np.sum(rec >= t) == 0:
p = 0
else:
p = np.max(prec[rec >= t])
ap = ap + p / 11.
else:
# correct AP calculation
# first append sentinel values at the end
mrec = np.concatenate(([0.], rec, [1.]))
mpre = np.concatenate(([0.], prec, [0.]))
# compute the precision envelope
for i in range(mpre.size - 1, 0, -1):
mpre[i - 1] = np.maximum(mpre[i - 1], mpre[i])
# to calculate area under PR curve, look for points
# where X axis (recall) changes value
i = np.where(mrec[1:] != mrec[:-1])[0]
# and sum (\Delta recall) * prec
ap = np.sum((mrec[i + 1] - mrec[i]) * mpre[i + 1])
return ap
def voc_eval(detpath, annopath, test_imgid_list, cls_name, ovthresh=0.5,
use_07_metric=False, use_diff=False):
# 1. parse xml to get gtboxes
# read list of images
imagenames = test_imgid_list
recs = {}
for i, imagename in enumerate(imagenames):
recs[imagename] = parse_rec(os.path.join(annopath, imagename + '.xml'))
# if i % 100 == 0:
# print('Reading annotation for {:d}/{:d}'.format(
# i + 1, len(imagenames)))
# 2. get gtboxes for this class.
class_recs = {}
num_pos = 0
# if cls_name == 'person':
# print ("aaa")
for imagename in imagenames:
R = [obj for obj in recs[imagename] if obj['name'] == cls_name]
bbox = np.array([x['bbox'] for x in R])
if use_diff:
difficult = np.array([False for x in R]).astype(np.bool)
else:
difficult = np.array([x['difficult'] for x in R]).astype(np.bool)
det = [False] * len(R)
num_pos = num_pos + sum(~difficult) # ignored the diffcult boxes
class_recs[imagename] = {'bbox': bbox,
'difficult': difficult,
'det': det} # det means that gtboxes has already been detected
# 3. read the detection file
detfile = os.path.join(detpath, "det_" + cls_name + ".txt")
with open(detfile, 'r') as f:
lines = f.readlines()
# for a line. that is [img_name, confidence, xmin, ymin, xmax, ymax]
splitlines = [x.strip().split(' ') for x in lines] # a list that include a list
image_ids = [x[0] for x in splitlines] # img_id is img_name
confidence = np.array([float(x[1]) for x in splitlines])
BB = np.array([[float(z) for z in x[2:]] for x in splitlines])
nd = len(image_ids) # num of detections. That, a line is a det_box.
tp = np.zeros(nd)
fp = np.zeros(nd)
if BB.shape[0] > 0:
# sort by confidence
sorted_ind = np.argsort(-confidence)
sorted_scores = np.sort(-confidence)
BB = BB[sorted_ind, :]
image_ids = [image_ids[x] for x in sorted_ind] # reorder the img_name
# go down dets and mark TPs and FPs
for d in range(nd):
R = class_recs[image_ids[d]] # img_id is img_name
bb = BB[d, :].astype(float)
ovmax = -np.inf
BBGT = R['bbox'].astype(float)
if BBGT.size > 0:
# compute overlaps
# intersection
ixmin = np.maximum(BBGT[:, 0], bb[0])
iymin = np.maximum(BBGT[:, 1], bb[1])
ixmax = np.minimum(BBGT[:, 2], bb[2])
iymax = np.minimum(BBGT[:, 3], bb[3])
iw = np.maximum(ixmax - ixmin + 1., 0.)
ih = np.maximum(iymax - iymin + 1., 0.)
inters = iw * ih
# union
uni = ((bb[2] - bb[0] + 1.) * (bb[3] - bb[1] + 1.) +
(BBGT[:, 2] - BBGT[:, 0] + 1.) *
(BBGT[:, 3] - BBGT[:, 1] + 1.) - inters)
overlaps = inters / uni
ovmax = np.max(overlaps)
jmax = np.argmax(overlaps)
if ovmax > ovthresh:
if not R['difficult'][jmax]:
if not R['det'][jmax]:
tp[d] = 1.
R['det'][jmax] = 1
else:
fp[d] = 1.
else:
fp[d] = 1.
# 4. get recall, precison and AP
fp = np.cumsum(fp)
tp = np.cumsum(tp)
rec = tp / float(num_pos)
# avoid divide by zero in case the first detection matches a difficult
# ground truth
prec = tp / np.maximum(tp + fp, np.finfo(np.float64).eps)
ap = voc_ap(rec, prec, use_07_metric)
return rec, prec, ap
def do_python_eval(test_imgid_list, test_annotation_path):
AP_list = []
# import matplotlib.pyplot as plt
# import matplotlib.colors as colors
# color_list = colors.cnames.keys()[::6]
for cls, index in NAME_LABEL_MAP.items():
if cls == 'back_ground':
continue
recall, precision, AP = voc_eval(detpath=os.path.join(cfgs.EVALUATE_DIR, cfgs.VERSION),
test_imgid_list=test_imgid_list,
cls_name=cls,
annopath=test_annotation_path,
use_07_metric=cfgs.USE_07_METRIC)
AP_list += [AP]
print("cls : {}|| Recall: {} || Precison: {}|| AP: {}".format(cls, recall[-1], precision[-1], AP))
# plt.plot(recall, precision, label=cls, color=color_list[index])
# plt.legend(loc='upper right')
# print(10*"__")
# plt.show()
# plt.savefig(cfgs.VERSION+'.jpg')
print("mAP is : {}".format(np.mean(AP_list)))
def voc_evaluate_detections(all_boxes, test_annotation_path, test_imgid_list):
'''
all_boxes:形状为(n,m,6)n表示有多少张图像，m表示一张图像有多少个检测物体，6表示 [category, score, xmin, ymin, xmax, ymax]
test_imgid_list：表示评估图片的图片名列表，其形状为（n,）
test_annotation_path：标签xml所在的文件夹
:param all_boxes: is a list. each item reprensent the detections of a img.The detections is a array. shape is [-1, 6]. [category, score, xmin, ymin, xmax, ymax].Note that: if none detections in this img. that the detetions is : []
:return:
'''
test_imgid_list = [item.split('.')[0] for item in test_imgid_list]
write_voc_results_file(all_boxes, test_imgid_list=test_imgid_list,
det_save_dir=os.path.join(cfgs.EVALUATE_DIR, cfgs.VERSION))
do_python_eval(test_imgid_list, test_annotation_path=test_annotation_path)
test_annotation_path = r"E:\data\yolov3\annotations"
import pickle
with open(r"E:\data\yolov3\test\evaluate\all_boxes.pkl", 'rb') as opened_file: # 读出文件
all_boxes = pickle.load(opened_file)
test_imgid_list = []
with open(r'E:\data\yolov3\test\evaluate\test_imgid_list.txt') as f:
lines = f.readlines()
for line in lines:
test_imgid_list.append(line.split("\n")[0])
voc_evaluate_detections(all_boxes, test_annotation_path, test_imgid_list)主なパラメータ
パラメータの最初の部分
これは主に一時ファイルを保存するためのもので、自分で設定できます。設定してもかまいません。
cfgs.EVALUATE_DIR = r "E：\ data \ yolov3 \ test \ evaluate"
cfgs.VERSION = r "cls_dir"
パラメータの2番目の部分
xmlファイルを保存するフォルダは次のとおりです。具体的なxml形式はvocに似ています。
test_annotation_path = r "E：\ data \ yolov3 \ annotations"
パラメータの3番目の部分
open（r "E：\ data \ yolov3 \ test \ evaluate \ all_boxes.pkl"、 'rb'）をopened_fileとして
このall_boxes.pklは一連の配列を保持し、特定の形状は（n、m、6）です。nは画像の数を表し、mは画像が持つ検出オブジェクトの数を表し、6は[category、score、xmin 、ymin、xmax、ymax]。
パラメータの4番目の部分
open（r'E：\ data \ yolov3 \ test \ evaluate \ test_imgid_list.txt '）をfとして
これは主にtest_imgid_list.txtファイルであり、主に.pngや他のサフィックスなしで画像の名前を保存し、パスなしで、それは単に
00001
00002
00003
等
注目
3番目の部分のパラメーターと4番目の部分のパラメーターには1対1の対応が必要です
