File size: 5,252 Bytes
1da285f | 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 | import os
import subprocess
from concurrent.futures import ThreadPoolExecutor
os.environ['split'] = 'genre'
MIN_APPS_PER_CAT = 5
split = os.getenv('split', '')
suffix_split = f'-{split}' if split else ''
GT = f'../gts/det/interactable/{split}.json'
RESULT_DIR = f'./results{suffix_split}'
RAW_RESULT_DIR = f'{RESULT_DIR}/raw'
GUIDANCES = [
# f'../results/ours/aaa_icse_test_set_merged_gpt4v_gpt4v_ape_d_object_bbox_r1i_bf.json',
# f'../results/ours/realfse_union3_claude35sonnet_ape_d_object_bbox_gpu0123_i_bf.json',
f'../results/ours/realfse_union3_gemini15pro_ape_d_object_bbox_gpu0123_i_bf.json',
# f'../results/ours/realfse_union3_gpt4v_ape_d_object_bbox_gpu0123_i_bf.json',
f'../results/baselines/{split}/det/interactable/YOLO.json',
# f'../results/baselines/{split}/det/interactable/GPT4V-E2E.json',
f'../results/baselines/{split}/det/interactable/internVL-E2E.json',
f'../results/baselines/{split}/det/interactable/Claude4_5-sonnet-E2E.json'
]
CAT_COMPARE_GUIDANCE = f'../results/ours/realfse_union3_gemini15pro_ape_d_object_bbox_gpu0123_i_bf.json'
BASELINE_GUIDANCE = f'../results/baselines/{split}/det/interactable/YOLO.json'
EVAL_UTILS_ROOT = './eval_utils'
GEN_INTERACT_SCRIPT = f'{EVAL_UTILS_ROOT}/gen_interact.py'
TIME_EVAL_SCRIPT = f'{EVAL_UTILS_ROOT}/time_evaluation.py'
CAT_EVAL_SCRIPT = f'{EVAL_UTILS_ROOT}/cat_evaluation.py'
AVG_SCRIPT = f'{EVAL_UTILS_ROOT}/time_avg_metric.py'
TIME_PLOT_SCRIPT = f'{EVAL_UTILS_ROOT}/time_plot.py'
CAT_PLOT_SCRIPT = f'{EVAL_UTILS_ROOT}/cat_plot.py'
TO_LATEX_SCRIPT = f'{EVAL_UTILS_ROOT}/cat_2latex.py'
ROUND = 30
os.environ['ROUND'] = str(ROUND)
os.makedirs(RAW_RESULT_DIR, exist_ok=True)
def generate_interact(inter_path, guide=None):
if not os.path.exists(inter_path + '.json'):
cli = f'python {GEN_INTERACT_SCRIPT} -a {GT} -o {inter_path + ".json"}'
if guide:
cli += f' -g {guide}'
print(cli)
subprocess.run(cli.split())
else:
print(f'{inter_path}.json already exists')
return inter_path
def time_eval_interact(inter_path, guide=None):
cli = f'python {TIME_EVAL_SCRIPT} -a {GT} -i {inter_path + ".json"} -o {inter_path + ".csv"}'
if guide:
cli += f' -p {guide}'
print(cli)
subprocess.run(cli.split())
return inter_path
def avg_metric(series):
cli = f'python {AVG_SCRIPT} -s {series}'
print(cli)
subprocess.run(cli.split())
def cat_eval(series, guide=None):
cli = f'python {CAT_EVAL_SCRIPT} -a {GT} -is {RAW_RESULT_DIR}/{series} -o {RESULT_DIR}/cat_eval/{series}'
if guide:
cli += f' -p {guide}'
print(cli)
subprocess.run(cli.split())
def cat_plot():
cli = (f'python {CAT_PLOT_SCRIPT} '
f'-r {RESULT_DIR}/cat_eval/{os.path.basename(BASELINE_GUIDANCE).split(".")[0]} '
f'-g {RESULT_DIR}/cat_eval/{os.path.basename(CAT_COMPARE_GUIDANCE).split(".")[0]} '
f'-t coverage_rate '
f'-m {MIN_APPS_PER_CAT}'
)
print(cli)
subprocess.run(cli.split())
# cli = f'python {CAT_PLOT_SCRIPT} ' + \
# f'-r {RESULT_DIR}/cat_eval/random ' + \
# f'-g {RESULT_DIR}/cat_eval/{os.path.basename(CAT_COMPARE_GUIDANCE).split(".")[0]} ' + \
# f'-t effective_interacts_rate'
# print(cli)
# subprocess.run(cli.split())
# cli = f'python {CAT_PLOT_SCRIPT} ' + \
# f'-r {RESULT_DIR}/cat_eval/YOLO ' + \
# f'-g {RESULT_DIR}/cat_eval/{os.path.basename(CAT_COMPARE_GUIDANCE).split(".")[0]} ' + \
# f'-t effective_interacts_cnt'
cli = (f'python {CAT_PLOT_SCRIPT} '
f'-r {RESULT_DIR}/cat_eval/{os.path.basename(BASELINE_GUIDANCE).split(".")[0]} '
f'-g {RESULT_DIR}/cat_eval/{os.path.basename(CAT_COMPARE_GUIDANCE).split(".")[0]} '
f'-t effective_interacts_cnt '
f'-m {MIN_APPS_PER_CAT}'
)
print(cli)
subprocess.run(cli.split())
def main():
with ThreadPoolExecutor(max_workers=16) as executor:
futures = []
for i in range(ROUND):
random_inter_path = os.path.join(RAW_RESULT_DIR, 'random' + str(i))
futures.append(executor.submit(generate_interact, random_inter_path))
for future in futures:
inter_path = future.result()
executor.submit(time_eval_interact, inter_path)
avg_metric('random')
cat_eval('random')
for guide in GUIDANCES:
with ThreadPoolExecutor(max_workers=16) as executor:
futures = []
for i in range(ROUND):
guide_inter_path = os.path.join(RAW_RESULT_DIR, os.path.basename(guide).split('.')[0] + str(i))
futures.append(executor.submit(generate_interact, guide_inter_path, guide))
for future in futures:
inter_path = future.result()
executor.submit(time_eval_interact, inter_path, guide)
avg_metric(os.path.basename(guide).split(".")[0])
cat_eval(os.path.basename(guide).split(".")[0], guide)
pass
cli = f'python {TIME_PLOT_SCRIPT}'
print(cli)
subprocess.run(cli.split())
cat_plot()
cli = f'python {TO_LATEX_SCRIPT} -gt {GT} -m {MIN_APPS_PER_CAT}'
print(cli)
subprocess.run(cli.split())
if __name__ == '__main__':
main()
|