-
Notifications
You must be signed in to change notification settings - Fork 0
/
inference.py
67 lines (52 loc) · 2 KB
/
inference.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
import os
import sys
import argparse
from random import seed
import numpy as np
import torch
from src.config import load_config, load_scene_list, load_invalid_frames
from src.vfuse import VFUSE
from src.tools.consensus_filtering import consensus_filtering
from src.evaluation.dtu.eval import dtu_point_eval
# argument parsing
parser = argparse.ArgumentParser(description="V-FUSE Network inference.")
parser.add_argument("--config_path", type=str, help="Configuration path.")
parser.add_argument("--dataset", type=str, help='Current dataset being used.', choices=["blendedmvs", "dtu", "tnt"], required=True)
ARGS = parser.parse_args()
def main():
#### Load Configuration ####
cfg = load_config(os.path.join(ARGS.config_path, f"{ARGS.dataset}.yaml"))
cfg["mode"] = "inference"
# set random seed
torch.manual_seed(cfg["seed"])
seed(cfg["seed"])
np.random.seed(cfg["seed"])
#### Load Scene Lists ####
scenes_list = os.path.join(cfg["scene_list_path"], "inference.txt")
with open(scenes_list,'r') as sf:
scenes = sf.readlines()
scenes = [s.strip() for s in scenes]
# metric used for evaluation
acc_sum = 0.0
comp_sum = 0.0
ovr_sum = 0.0
for scene in scenes:
print(f"\n----Running V-FUSE on {scene}----")
#### INFERENCE ####
pipeline = VFUSE(cfg, inference_scene=[scene])
pipeline.inference()
consensus_filtering(cfg, scene, pipeline.dataset.get_cluster_list_file(scene), f"{scene}.ply")
#### EVALUATION ####
if cfg["eval"]["run_eval"]:
if (ARGS.dataset == "dtu"):
acc, comp, ovr, prec, rec = dtu_point_eval(cfg, scene)
acc_sum += acc
comp_sum += comp
ovr_sum += ovr
if cfg["eval"]["run_eval"]:
print("\n---Total Results---")
print(f"Acc: {acc_sum/len(scenes):0.3f}")
print(f"Comp: {comp_sum/len(scenes):0.3f}")
print(f"Ovr: {ovr_sum/len(scenes):0.3f}")
if __name__ == '__main__':
main()