app.py

from transformers import SegformerForSemanticSegmentation
from transformers import SegformerImageProcessor
from PIL import Image
import gradio as gr  
import numpy as np
import random 
import cv2
import torch

from imutils import perspective


def midpoint(ptA, ptB):
	return ((ptA[0] + ptB[0]) * 0.5, (ptA[1] + ptB[1]) * 0.5)
# Load in image, convert to gray scale, and Otsu's threshold
kernel1 =( np.ones((5,5), dtype=np.float32))
blur_radius=0.5
kernel_sharpening = np.array([[-1,-1,-1], 
                              [-1,9,-1], 
                             [-1,-1,-1]])*(1/9)


def cca_analysis(image,predicted_mask):

    image2=np.asarray(image)
    print(image.shape)
    image = cv2.resize(predicted_mask, (image2.shape[1],image2.shape[1]), interpolation = cv2.INTER_AREA)

    image=cv2.morphologyEx(image, cv2.MORPH_OPEN, kernel1,iterations=1 )
    gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
    thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]

    labels=cv2.connectedComponents(thresh,connectivity=8)[1]       
    a=np.unique(labels)
    count2=0
    for label in a:
        if label == 0:
            continue

        # Create a mask
        mask = np.zeros(thresh.shape, dtype="uint8")
        mask[labels == label] = 255

        # Find contours and determine contour area
        cnts,hieararch = cv2.findContours(mask.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
        cnts = cnts[0]
        c_area = cv2.contourArea(cnts)
        # threshhold for tooth count
        if c_area>100:
            count2+=1
        
        rect = cv2.minAreaRect(cnts)
        box = cv2.boxPoints(rect)
        box = np.array(box, dtype="int")    
        box = perspective.order_points(box)
        color1 = (list(np.random.choice(range(150), size=3)))  
        color =[int(color1[0]), int(color1[1]), int(color1[2])]  
        cv2.drawContours(image2,[box.astype("int")],0,color,2)
        (tl,tr,br,bl)=box
        
        (tltrX,tltrY)=midpoint(tl,tr)
        (blbrX,blbrY)=midpoint(bl,br)
        # compute the midpoint between the top-left and top-right points,
        # followed by the midpoint between the top-righ and bottom-right
        (tlblX,tlblY)=midpoint(tl,bl)
        (trbrX,trbrY)=midpoint(tr,br)
        # draw the midpoints on the image
        cv2.circle(image2, (int(tltrX), int(tltrY)), 5, (255, 0, 0), -1)
        cv2.circle(image2, (int(blbrX), int(blbrY)), 5, (255, 0, 0), -1)
        cv2.circle(image2, (int(tlblX), int(tlblY)), 5, (255, 0, 0), -1)
        cv2.circle(image2, (int(trbrX), int(trbrY)), 5, (255, 0, 0), -1)
        cv2.line(image2, (int(tltrX), int(tltrY)), (int(blbrX), int(blbrY)),color, 2)
        cv2.line(image2, (int(tlblX), int(tlblY)), (int(trbrX), int(trbrY)),color, 2)
    return image2

def to_rgb(img):
    result_new=np.zeros((img.shape[1],img.shape[0],3))
    result_new[:,:,0]=img
    result_new[:,:,1]=img
    result_new[:,:,2]=img
    result_new=np.uint8(result_new*255)
    return result_new

image_list = [
    "data/1.png",
    "data/2.png",
    "data/3.png",
    "data/4.png",
]

model_path = ['deprem-ml/deprem_satellite_semantic_whu']

def visualize_instance_seg_mask(mask):
    # Initialize image with zeros with the image resolution
    # of the segmentation mask and 3 channels
    image = np.zeros((mask.shape[0], mask.shape[1], 3))

    # Create labels
    labels = np.unique(mask)
    label2color = {
        label: (
            random.randint(0, 255),
            random.randint(0, 255),
            random.randint(0, 255),
        )
        for label in labels
    }

    for height in range(image.shape[0]):
        for width in range(image.shape[1]):
            image[height, width, :] = label2color[mask[height, width]]

    image = image / 255
    return image


def Segformer_Segmentation(image_path, model_id,postpro): 
    output_save = "output.png"
    
    test_image = cv2.imread(image_path)

    model = SegformerForSemanticSegmentation.from_pretrained(model_id)
    proccessor = SegformerImageProcessor(model_id)
    
    inputs = proccessor(images=test_image, return_tensors="pt")
    with torch.no_grad():
        outputs = model(**inputs)
        
    result = proccessor.post_process_semantic_segmentation(outputs)[0]
    result = np.array(result)
    if postpro=="Connected Components Labelling":
        result=to_rgb(result)
        result=cca_analysis(test_image,result)
    else:
        result = visualize_instance_seg_mask(result)
        result=result*255

    cv2.imwrite(output_save, result)
    return image_path, output_save

examples = [[image_list[0], "deprem-ml/deprem_satellite_semantic_whu"],
            [image_list[1], "deprem-ml/deprem_satellite_semantic_whu"],
            [image_list[2], "deprem-ml/deprem_satellite_semantic_whu"],
            [image_list[3], "deprem-ml/deprem_satellite_semantic_whu"]]

title = "Deprem ML - Segformer Semantic Segmentation"

app = gr.Blocks()
with app:
    gr.HTML("<h1 style='text-align: center'>{}</h1>".format(title))    
    with gr.Row():
        with gr.Column():
            input_video = gr.Image(type='filepath')
            model_id =  gr.Dropdown(value=model_path[0], choices=model_path,label="Model Name")
            cca =  gr.Dropdown(value="Connected Components Labelling", choices=["Connected Components Labelling","No Post Process"],label="Post Process")

            input_video_button = gr.Button(value="Predict")    
                    
        with gr.Column():
            output_orijinal_image = gr.Image(type='filepath')
            
        with gr.Column():
            output_mask_image =  gr.Image(type='filepath')

        
    gr.Examples(examples, inputs=[input_video, model_id,cca], outputs=[output_orijinal_image, output_mask_image], fn=Segformer_Segmentation, cache_examples=True)
    input_video_button.click(Segformer_Segmentation, inputs=[input_video, model_id,cca], outputs=[output_orijinal_image, output_mask_image])

app.launch(debug=True)