forked from salaniz/pycocoevalcap
-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy patheval.py
85 lines (76 loc) · 3.08 KB
/
eval.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
__author__ = 'tylin'
from .tokenizer.ptbtokenizer import PTBTokenizer
from .bleu.bleu import Bleu
from .meteor.meteor import Meteor
from .rouge.rouge import Rouge
from .cider.cider import Cider
from .spice.spice import Spice
from .clipscore.clipscore import ClipScore
class COCOEvalCap:
def __init__(self, coco, cocoRes):
self.evalImgs = []
self.eval = {}
self.imgToEval = {}
self.coco = coco
self.cocoRes = cocoRes
self.params = {'image_id': coco.getImgIds()}
def evaluate(self):
imgIds = self.params['image_id']
# imgIds = self.coco.getImgIds()
gts = {}
res = {}
for imgId in imgIds:
gts[imgId] = self.coco.imgToAnns[imgId]
res[imgId] = self.cocoRes.imgToAnns[imgId]
# =================================================
# Set up scorers
# =================================================
print('tokenization...')
tokenizer = PTBTokenizer()
gts = tokenizer.tokenize(gts)
res = tokenizer.tokenize(res)
# =================================================
# Set up scorers
# =================================================
print('setting up scorers...')
scorers = [
(Bleu(4), ["Bleu_1", "Bleu_2", "Bleu_3", "Bleu_4"]),
(Meteor(),"METEOR"),
(Rouge(), "ROUGE_L"),
(Cider(), "CIDEr"),
(Spice(), "SPICE"),
(ClipScore(), ["CLIPScore", "RefCLIPScore"])
]
# =================================================
# Compute scores
# =================================================
for scorer, method in scorers:
print('computing %s score...'%(scorer.method()))
score, scores = scorer.compute_score(gts, res)
if type(method) == list:
for sc, m in zip(score, method):
self.setEval(sc, m)
if "CLIP" in m:
scores_clip = [i[m] for i in scores]
self.setImgToEvalImgs(scores_clip, gts.keys(), m)
elif 'Bleu' in m:
scores_bleu = scores[method.index(m)]
self.setImgToEvalImgs(scores_bleu, gts.keys(), m)
else:
self.setImgToEvalImgs(scores, gts.keys(), m)
print("%s: %0.3f"%(m, sc))
else:
self.setEval(score, method)
self.setImgToEvalImgs(scores, gts.keys(), method)
print("%s: %0.3f"%(method, score))
self.setEvalImgs()
def setEval(self, score, method):
self.eval[method] = score
def setImgToEvalImgs(self, scores, imgIds, method):
for imgId, score in zip(imgIds, scores):
if not imgId in self.imgToEval:
self.imgToEval[imgId] = {}
self.imgToEval[imgId]["image_id"] = imgId
self.imgToEval[imgId][method] = score
def setEvalImgs(self):
self.evalImgs = [eval for imgId, eval in self.imgToEval.items()]