File size: 1,425 Bytes
0093fb3
 
3b4a7c0
a35e152
7efb1aa
b06f7bd
7efb1aa
1204185
 
7efb1aa
1204185
 
 
62d8c85
7efb1aa
1204185
0093fb3
884c19e
 
 
 
ad83853
33afddf
 
7efb1aa
 
 
 
 
 
 
 
 
 
 
05958ce
7efb1aa
 
33afddf
7efb1aa
0093fb3
 
25c8f14
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
import gradio as gr
import numpy as np
import flip_evaluator as flip

def compute_simple_diff(a: np.array, b: np.array) -> (np.array, float):
    '''compute difference of two arrays'''
    return a - b, str(np.round(np.mean(a-b), 4))


def compute_flip_diff(a: np.array, b: np.array) -> (np.array, float):
    '''
    compute NVIDIA nvlab's FLIP difference between two image arrays
    '''
    error_map, mean_error, _ = flip.evaluate(a, b, "HDR")
    return error_map, str(np.round(mean_error,4))
    

examples = [
    ["examples/Office_ShadowAcne/reference.png",
    "examples/Office_ShadowAcne/test.png"]
            ]

with gr.Blocks() as demo:
    gr.Markdown("Compute difference using NVIDIA's [FLIP](https://github.com/NVlabs/flip/)")
    with gr.Row(equal_height=True):
        # define inputs and outputs
        before = gr.Image(label="Before")
        after = gr.Image(label="After")
        diff = gr.Image(label="Difference")
    with gr.Row(equal_height=True):
        with gr.Column(scale=2):          
            button = gr.Button("Compute difference")
        with gr.Column(scale=1, min_width=200):
            meanbox = gr.Textbox(label="Mean difference")
    # interaction
    button.click(fn=compute_flip_diff,
                 inputs=[before, after],
                 outputs=[diff, meanbox]
                )
    gr.Examples(examples, [before, after])

if __name__ == "__main__":
    demo.launch()