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(
    "nickmuchi/segformer-b4-finetuned-segments-sidewalk"
)
model = TFSegformerForSemanticSegmentation.from_pretrained(
    "nickmuchi/segformer-b4-finetuned-segments-sidewalk",
    from_pt=True
)

def ade_palette():
    """ADE20K palette that maps each class to RGB values."""
    return [
        [56, 123, 45],
        [45, 189, 106],
        [210, 32, 89],
        [0, 0, 0],
        [56, 45, 189],
        [234, 123, 67],
        [155, 102, 200],
        [33, 147, 176],
        [255, 255, 255],
        [67, 123, 89],
        [190, 60, 45],
        [134, 112, 200],
        [78, 56, 123],
        [200, 56, 123],
        [87, 92, 204],
        [120, 56, 123],
        [45, 78, 123],
        [112, 185, 212],
        [156, 200, 56],
        [32, 90, 210],
        [56, 123, 67],
        [180, 56, 123],
        [123, 67, 45],
        [45, 134, 200],
        [67, 56, 123],
        [78, 123, 67],
        [32, 210, 90],
        [45, 56, 189],
        [123, 56, 123],
        [56, 156, 200],
        [189, 56, 45],
        [112, 200, 56],
        [204, 87, 92],
        [200, 32, 90],
        [123, 45, 78],
    ]

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 on_mouseover(mouse_info):
    x, y = mouse_info["position"]
    
    # 사용자 정의 함수를 호출하여 특정 영역을 강조
    highlight_segmentation(x, y)

def highlight_segmentation(x, y):
    # 마우스 위치에 대한 세그멘테이션 정보 가져오기
    seg_value = seg.numpy()[y, x]

    # 세그멘테이션 정보 출력
    print(f"Mouse over position: ({x}, {y}), Segmentation value: {seg_value}")
    
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.
    global seg 
    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'],
                    title="Gradio 기반의 Hugging Face 영상 분할 모델 데모 구현 과제 3",
                    description="차도, 인도, 자전거도로 등의 사진만 넣어주세요",
                    examples=["example1.jpg", "example2.jpg", "example3.jpeg", "example4.jpg"],
                    allow_flagging='never',
                    events=[gr.Mouse(on_mouseover)])

demo.launch()