app.py

import gradio as gr

from matplotlib import gridspec
import matplotlib.pyplot as plt
import numpy as np
from PIL import Image
import tensorflow as tf
from transformers import SegformerFeatureExtractor, TFSegformerForSemanticSegmentation

feature_extractor = SegformerFeatureExtractor.from_pretrained(
    "nvidia/segformer-b1-finetuned-cityscapes-1024-1024"
)
model = TFSegformerForSemanticSegmentation.from_pretrained(
    "nvidia/segformer-b1-finetuned-cityscapes-1024-1024"
)

def ade_palette():
    """ADE20K palette that maps each class to RGB values."""
    return [
        [255, 0, 0],  # Class 0
        [0, 255, 0],  # Class 1
        [0, 0, 255],  # Class 2
        [255, 255, 0],  # Class 3
        [255, 0, 255],  # Class 4
        [0, 255, 255],  # Class 5
        [128, 0, 0],  # Class 6
        [0, 128, 0],  # Class 7
        [0, 0, 128],  # Class 8
        [128, 128, 0],  # Class 9
        [128, 0, 128],  # Class 10
        [0, 128, 128],  # Class 11
        [64, 0, 0],  # Class 12
        [0, 64, 0],  # Class 13
        [0, 0, 64],  # Class 14
        [64, 64, 0],  # Class 15
        [64, 0, 64],  # Class 16
        [0, 64, 64],  # Class 17
        [64, 64, 64],

    ]

labels_list = []

with open(r'labels.txt', 'r') as fp:
    for line in fp:
        labels_list.append(line[:-1])

colormap = np.asarray(ade_palette())

def label_to_color_image(label):
    if label.ndim != 2:
        raise ValueError("Expect 2-D input label")

    if np.max(label) >= len(colormap):
        raise ValueError("label value too large.")
    return colormap[label]

def draw_plot(pred_img, seg):
    fig = plt.figure(figsize=(20, 15))

    grid_spec = gridspec.GridSpec(1, 2, width_ratios=[6, 1])

    plt.subplot(grid_spec[0])
    plt.imshow(pred_img)
    plt.axis('off')
    LABEL_NAMES = np.asarray(labels_list)
    FULL_LABEL_MAP = np.arange(len(LABEL_NAMES)).reshape(len(LABEL_NAMES), 1)
    FULL_COLOR_MAP = label_to_color_image(FULL_LABEL_MAP)

    unique_labels = np.unique(seg.numpy().astype("uint8"))
    ax = plt.subplot(grid_spec[1])
    plt.imshow(FULL_COLOR_MAP[unique_labels].astype(np.uint8), interpolation="nearest")
    ax.yaxis.tick_right()
    plt.yticks(range(len(unique_labels)), LABEL_NAMES[unique_labels])
    plt.xticks([], [])
    ax.tick_params(width=0.0, labelsize=25)
    return fig

def sepia(input_img):
    input_img = Image.fromarray(input_img)

    inputs = feature_extractor(images=input_img, return_tensors="tf")
    outputs = model(**inputs)
    logits = outputs.logits

    logits = tf.transpose(logits, [0, 2, 3, 1])
    logits = tf.image.resize(
        logits, input_img.size[::-1]
    )  # We reverse the shape of `image` because `image.size` returns width and height.
    seg = tf.math.argmax(logits, axis=-1)[0]

    color_seg = np.zeros(
        (seg.shape[0], seg.shape[1], 3), dtype=np.uint8
    )  # height, width, 3
    for label, color in enumerate(colormap):
        color_seg[seg.numpy() == label, :] = color

    # Show image + mask
    pred_img = np.array(input_img) * 0.5 + color_seg * 0.5
    pred_img = pred_img.astype(np.uint8)

    fig = draw_plot(pred_img, seg)
    return fig

# demo = gr.Interface(fn=sepia,
#                     inputs=gr.Image(shape=(400, 600)),
#                     outputs=['plot'],
#                     examples=["22.jpg", "33.jpg"],
#                     allow_flagging='never')

demo = gr.Interface(
    fn=sepia,
    inputs=gr.Image(shape=(400, 600)),
    outputs='plot',  # 'plot' 대신 'image'도 사용 가능
    examples=["22.jpg", "33.jpg"],
    allow_flagging='never',
    title="201912064 Park MinGyu",
    description="sample image - city road",
    theme="huggingface",
    live=True,  # 실시간 업데이트 활성화
    interpretation="default",  # 또는 'none'으로 설정하여 해석 기능 비활성화
    layout="horizontal",  # 'horizontal' 또는 'vertical'로 설정 가능
)


demo.launch()