forked from MediaTek-NeuroPilot/mai21-learned-smartphone-isp
-
Notifications
You must be signed in to change notification settings - Fork 0
/
inference_tflite.py
105 lines (85 loc) · 3.52 KB
/
inference_tflite.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
#################################
# Inferening using TFLite model #
#################################
import os
from os import path
import glob
import itertools
import time
import argparse
import numpy as np
import imageio
import tqdm
import math
import tensorflow as tf
from load_dataset import extract_bayer_channels
def preprocess(image_path):
I = np.asarray(imageio.imread(image_path))
I = extract_bayer_channels(I)
I = I[0:256//2, 0:256//2, :]
I = np.reshape(I, [1, I.shape[0], I.shape[1], 4])
return I
def psnr(x, y):
diff = x.astype(np.float32) - y.astype(np.float32)
mse = np.mean(diff**2)
ret = 20 * math.log10(1.0/math.sqrt(mse))
return ret
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--dataset_dir', type=str, default='raw_images')
parser.add_argument('--dir_type', type=str, default='test')
parser.add_argument('--phone_dir', type=str, default='mediatek_raw')
parser.add_argument('--dslr_dir', type=str, default=None)
parser.add_argument('--model_file', type=str, default='models/original/punet_pretrained.tflite')
parser.add_argument('--save_results', action='store_true')
parser.add_argument('--save_dir', type=str, default='results')
args = parser.parse_args()
# initialization of TFLite interpreter
interpreter = tf.lite.Interpreter(
model_path=args.model_file,
)
interpreter.allocate_tensors()
input_details = interpreter.get_input_details()
output_details = interpreter.get_output_details()
# check the type of the input tensor
floating_model = input_details[0]['dtype'] == np.float32
# set input directory
input_dir = path.join(args.dataset_dir, args.dir_type, args.phone_dir)
target_dir = path.join(args.dataset_dir, args.dir_type, args.dslr_dir) if args.dslr_dir is not None else None
save_dir = path.join(args.save_dir, args.dir_type)
if args.save_results:
os.makedirs(save_dir, exist_ok=True)
# prepare data list
scan = lambda x, y: glob.glob(path.join(x, y))
input_list = [scan(input_dir, d) for d in os.listdir(input_dir)]
input_list = sorted([it for it in itertools.chain(*input_list)])
if target_dir is None:
target_list = [None for _ in input_list]
else:
target_list = [scan(target_dir, d) for d in os.listdir(target_dir)]
target_list = sorted([it for it in itertools.chain(*target_list)])
# pass each input image to the TFLite model
psnr_sum = 0.0
tq = tqdm.tqdm(zip(input_list, target_list), total=len(input_list))
for input_path, target_path in tq:
# print(input_path)
input_img = preprocess(input_path)
if target_path is not None:
target_img = np.asarray(imageio.imread(target_path))
target_img = target_img.astype(np.float32) / 255
interpreter.set_tensor(input_details[0]['index'], input_img)
interpreter.invoke()
output_img = interpreter.get_tensor(output_details[0]['index'])
output_img = np.clip(output_img, 0., 1.)
output_img = np.squeeze(output_img)
# calculate PSNR if ground truth is available
if target_path is not None:
psnr_sum += psnr(output_img, target_img)
# save results
if args.save_results:
save_as = input_path.replace(input_dir, save_dir)
imageio.imwrite(save_as, (output_img*255).astype(np.uint8))
if target_path is not None:
print('Avg. PSNR: {:.2f}'.format(psnr_sum / len(input_list)))
if __name__ == '__main__':
main()