File size: 1,449 Bytes
792a6ee |
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 |
result_model = '/proj/cvl/users/x_fahkh2/UI-R1/ckpt/Qwen2.5-VL-3B-Instruct-Trained'
tasks = ['mobile', 'desktop', 'web']
#tasks = ['desktop']
types = ['icon', 'text']
import os
import json
import re
result = {}
total = {}
for task in tasks:
ref_json = f'/proj/cvl/users/x_fahkh2/UI-R1/data/ScreenSpot/annotations/screenspot_{task}.json'
data = json.load(open(ref_json, 'r'))
bbox_type = {}
for d in data:
bbox = str(d['bbox']).strip('[]')
bbox_type[bbox] = d['data_type']
result_jsonl = os.path.join(result_model, f'infer/prediction_results_screenspot_{task}.jsonl')
with open(result_jsonl, 'r') as f:
lines = f.readlines()
for line in lines:
gt_bbox_match = re.search(r'"gt_bbox": \[(.*?)\]', line)
gt_bbox = gt_bbox_match.group(1)
# print(gt_bbox)
type = bbox_type[gt_bbox]
if "true" in line or "True" in line:
if f'{task}_{type}' not in result:
result[f'{task}_{type}'] = 1
else:
result[f'{task}_{type}'] = result[f'{task}_{type}'] + 1
if f'{task}_{type}' not in total:
total[f'{task}_{type}'] = 1
else:
total[f'{task}_{type}'] = total[f'{task}_{type}'] + 1
#print(result)
#print(total)
for k in result.keys():
print(k,result[k]/total[k])
print(sum(result.values()) / sum(total.values()))
|