-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy patheval_refseg.py
166 lines (142 loc) · 7.71 KB
/
eval_refseg.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
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
import os
import os.path as osp
import re
import cv2
import json
import argparse
from collections import defaultdict
import random
import numpy as np
from datetime import datetime
from PIL import Image
from tqdm import tqdm
import torch
from torch.utils.data import DataLoader
from minigpt4.common.config import Config
from minigpt4.common.eval_utils import prepare_texts, init_model, eval_parser, computeIoU
from minigpt4.conversation.conversation import CONV_VISION_minigptv2
from minigpt4.datasets.datasets.coco_seg_dataset import RefCOCOSegEvalData
from minigpt4.metrics.parsing_metric import HumParEvaluator
def list_of_str(arg):
return list(map(str, arg.split(',')))
parser = eval_parser()
parser.add_argument("--dataset", type=list_of_str, default='refcoco', help="dataset to evaluate")
parser.add_argument("--res", type=float, default=100.0, help="resolution used in refcoco")
parser.add_argument("--resample", action='store_true', help="resolution used in refcoco")
args = parser.parse_args()
cfg = Config(args)
eval_dict = {'refcoco': ['val','testA','testB'],
'refcoco+': ['val','testA','testB'],
'refcocog': ['val','test']}
model, vis_processor = init_model(args)
model.eval()
CONV_VISION = CONV_VISION_minigptv2
conv_temp = CONV_VISION.copy()
conv_temp.system = ""
#
model.eval()
save_path = cfg.run_cfg.save_path
timestamp = datetime.now().strftime('%Y-%m-%d_%H-%M-%S')
save_path = os.path.join(save_path, timestamp)
for dataset in args.dataset:
for split in eval_dict[dataset]:
eval_file_path = cfg.evaluation_datasets_cfg[dataset]["eval_file_path"]
img_path = cfg.evaluation_datasets_cfg[dataset]["img_path"]
batch_size = cfg.evaluation_datasets_cfg[dataset]["batch_size"]
max_new_tokens = cfg.evaluation_datasets_cfg[dataset]["max_new_tokens"]
with open(os.path.join(eval_file_path,f"{dataset}/{dataset}_seg_{split}.json"), 'r') as f:
refcoco = json.load(f)
# refcoco = refcoco[:10] # sample mini-set for quick evaluation
data = RefCOCOSegEvalData(refcoco, vis_processor, img_path)
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
minigpt4_predict_dict = defaultdict(dict) #defaultdict(list)
resamples = []
for images, questions, img_ids in tqdm(eval_dataloader):
texts = prepare_texts(questions, conv_temp) # warp the texts with conversation template
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
for answer, img_id, question in zip(answers, img_ids, questions):
answer = answer.replace("<unk>","").replace(" ","").strip()
pattern = r'\{(<\d{1,3}>)+\}'
if re.match(pattern, answer):
exp = question.replace("[refer segmentation] give me the boundary points of ","")
minigpt4_predict_dict[img_id][exp] = answer
#minigpt4_predict[img_id].append(answer)
else:
# print(question.replace('[refer segmentation] give me the boundary points of','').strip())
resamples.append({'img_id': img_id, 'sents': [question.replace('[refer segmentation] give me the boundary points of','').strip()]})
if args.resample:
for i in range(20):
data = RefCOCOSegEvalData(resamples, vis_processor, img_path)
resamples = []
eval_dataloader = DataLoader(data, batch_size=batch_size, shuffle=False)
for images, questions, img_ids in tqdm(eval_dataloader):
texts = prepare_texts(questions, conv_temp) # warp the texts with conversation template
answers = model.generate(images, texts, max_new_tokens=max_new_tokens, do_sample=False)
for answer, img_id, question in zip(answers, img_ids, questions):
answer = answer.replace("<unk>","").replace(" ","").strip()
pattern = r'\{<\d{1,3}><\d{1,3}><\d{1,3}><\d{1,3}>\}'
if re.match(pattern, answer) or i == 4:
exp = question.replace("[refer segmentation] give me the boundary points of ","")
minigpt4_predict_dict[img_id][exp] = answer
else:
# print(question.replace('[refer segmentation] give me the boundary points of','').strip())
resamples.append({'img_id': img_id, 'sents': [question.replace('[refer segmentation] give me the boundary points of','').strip()]})
if len(resamples) == 0:
break
file_save_path = os.path.join(save_path,f"{dataset}_{split}.json")
directory = os.path.dirname(file_save_path)
# 如果目录不存在，则创建它
if not os.path.exists(directory):
os.makedirs(directory, exist_ok=True)
vis_root = os.path.join(save_path, split, 'vis')
os.makedirs(vis_root, exist_ok=True)
std_dict = dict(minigpt4_predict_dict)
with open(file_save_path,'w') as f:
json.dump(std_dict, f)
gt_masks = []
pred_masks = []
res=args.res
evaluator = HumParEvaluator(class_names=[0, 1])
evaluator.reset()
for item in refcoco:
#refcoco_dict[item['img_id']] = item
img_id = item['img_id']
sent_id = item['sent_id']
show_image = cv2.imread(osp.join(img_path, img_id))
gt_segmentation = item['segmentation']
#item = refcoco_dict[img_id]
output = minigpt4_predict_dict[img_id].get(item['sents'], None)
if output is None:
# print(item['sents'])
output = '{<0><0><0><0>}'
height = item['height']
width = item['width']
seg_strs = output.split('<delim>')
seg_pattern = r"<(\d{1,3})>"
pred_segmentation = []
for seg in seg_strs:
poly_list = re.findall(seg_pattern, seg)
poly_list = poly_list[:-1] if len(poly_list) % 2 == 1 else poly_list[:] # TODO: how to deal with odd number of outputs
polygon = np.array(poly_list).reshape(-1, 2).astype(float)
polygon = polygon / res * np.array([width, height])
polygon_list = np.round(polygon, 2).flatten().tolist()
pred_segmentation.append(polygon_list)
gt_mask = np.zeros((height, width), dtype=np.uint8)
for seg in gt_segmentation:
polygon = np.array(seg).reshape(-1, 2)
cv2.fillConvexPoly(gt_mask, np.int32([polygon]), 1)
cv2.polylines(show_image, np.int32([polygon]), True, (0, 0, 255), 2)
gt_masks.append(gt_mask)
pred_mask = np.zeros_like(gt_mask, dtype=np.uint8)
for seg in pred_segmentation:
polygon = np.array(seg).reshape(-1, 2)
cv2.fillConvexPoly(pred_mask, np.int32([polygon]), 1)
cv2.polylines(show_image, np.int32([polygon]), True, (0, 255, 0), 1)
pred_masks.append(pred_mask)
save_file = osp.join(vis_root, img_id[:-4] + '_' + str(sent_id) + '.jpg')
cv2.imwrite(save_file, show_image)
evaluator.process(gt_masks, pred_masks)
print(f'Evalute results on {dataset}-{split}:')
results = evaluator.evaluate()
# CUDA_VISIBLE_DEVICES=5 python -m torch.distributed.run --master-port 29505 --nproc_per_node 1 eval_scripts/eval_ref.py \
# --cfg-path /home/human/codes/human/MiniGPT-4/eval_configs/minigptv2_benchmark_evaluation.yaml --dataset refcoco,refcoco+,refcocog #--resample