app.py

import os
import shutil
import cv2
import zipfile
import uuid

import numpy as np
import gradio as gr

from naming import im2c
from collections import Counter


COLOR_NAME = ['black', 'brown', 'blue', 'gray', 'green', 'orange', 'pink', 'purple', 'red', 'white', 'yellow']


def get_top_names(img):
    # resize images to smaller size
    anchor = 256
    width = img.shape[1]
    height = img.shape[0]
    if width > 512 or height > 512:
        if width >= height:
            dim = (np.floor(width/height*anchor).astype(int), anchor)
        else:
            dim = (anchor, np.floor(height/width*anchor).astype(int))
        img = cv2.resize(img, dim, interpolation=cv2.INTER_LINEAR)

    # obtain color names of all the pixels
    w2c = np.load('w2c11_j.npy').astype(np.float16)
    _, _, name_idx_img, _ = im2c(img, w2c)

    # compute the order of each name based on the numbers of each name
    filtered_counts = Counter(name_idx_img[name_idx_img <= 10])
    sorted_counts = sorted(filtered_counts.items(), key=lambda x: x[1], reverse=True)
    top_3_values = [num for num, count in sorted_counts[:3]]
    top_3_counts = [count/(dim[0]*dim[1])  for num, count in sorted_counts[:3]]
    top_3_colors = [COLOR_NAME[i] for i in top_3_values]
    # print("Top 3 colors:", top_3_counts)
    return top_3_values, top_3_counts, top_3_colors


def classify_and_log(images):
    # output_folder = "classified_results"
    # os.makedirs(output_folder, exist_ok=True)

     # create a temporary directory
    session_id = str(uuid.uuid4())
    output_dir = f"temp_{session_id}"
    os.makedirs(output_dir, exist_ok=True)


    category_folders = {i: os.path.join(output_dir, COLOR_NAME[i]) for i in range(11)}
    for folder in category_folders.values():
        os.makedirs(folder, exist_ok=True)

    log_file = os.path.join(output_dir, "top3colors.txt")
    
    results = {i: [] for i in range(11)}


    with open(log_file, "w") as log:
        for id_img, img in enumerate(images):
            filename = os.path.basename(img.name)
            img_array = cv2.imread(img).astype(np.float32)
            img_array = cv2.cvtColor(img_array, cv2.COLOR_BGR2RGB)

            cat_id, cat_counts, category = get_top_names(img_array)

            for i in range(3):
                if cat_counts[i] > 0.15:
                    target_path = os.path.join(category_folders[cat_id[i]], filename)
                    cv2.imwrite(target_path, cv2.cvtColor(img_array, cv2.COLOR_BGR2RGB))

            # print(f"Image:{filename} -> Top 3 colors:{category}\n")

            log.write(f"{filename} -> 1 {category[0]} {100*cat_counts[0]:.2f}%, 2 {category[1]} {100*cat_counts[1]:.2f}%, 3 {category[2]} {100*cat_counts[2]:.2f}%\n")

            results[cat_id[0]].append(target_path)


    # compile all images into a zip file
    zip_path = f"{output_dir}.zip"
    with zipfile.ZipFile(zip_path, "w", zipfile.ZIP_DEFLATED) as zipf:
        for root, _, files in os.walk(output_dir):
            for file in files:
                file_path = os.path.join(root, file)
                arcname = os.path.relpath(file_path, start=output_dir)
                zipf.write(file_path, arcname)

    # optional: clean up the output directory
    shutil.rmtree(output_dir)

    return zip_path


def swap_to_gallery(images):
    return gr.update(value=images, visible=True), gr.update(visible=True), gr.update(visible=False)

def upload_example_to_gallery(images, prompt, style, negative_prompt):
    return gr.update(value=images, visible=True), gr.update(visible=True), gr.update(visible=False)

def remove_back_to_files():
    return gr.update(visible=False), gr.update(visible=False), gr.update(visible=True)



with gr.Blocks() as demo:
    gr.Markdown("## Image color categorization")
    gr.Markdown("We categorize images into different classes based on the frequency of different colors appearing in each image.")
    gr.Markdown("The 11 color catergories include: black, brown, blue, gray, green, orange, pink, purple, red, white and yellow.")
    gr.Markdown("The classification is based on the color naming model from paper _Van De Weijer, Joost, et al. Learning color names for real-world applications. IEEE Transactions on Image Processing 18.7 (2009): 1512-1523._")
    gr.Markdown("The output results are in a zip file with all the images in the correspoding folders.")
    gr.Markdown("Note that one image can be classified into multiple categories (top 3 categories and frequency > 15%), and the top 3 categories are listed in the log file.")


    with gr.Row():
        with gr.Column():
            image_input = gr.File(
                        label="Drag/Select more than one images",
                        file_types=["image"],
                        file_count="multiple"
                    )
            uploaded_files = gr.Gallery(label="Your images", visible=False, columns=5, rows=1, height=200)


            with gr.Column(visible=False) as clear_button:
                remove_and_reupload = gr.ClearButton(value="Remove and upload new ones", components=image_input, size="sm")


        image_input.upload(fn=swap_to_gallery, inputs=image_input, outputs=[uploaded_files, clear_button, image_input])
        remove_and_reupload.click(fn=remove_back_to_files, outputs=[uploaded_files, clear_button, image_input])

    classify_btn = gr.Button("submit")

    # with gr.Row():
    #     image_output = {str(i): gr.Gallery(label=f"{i}") for i in range(11)}
    
    log_output = gr.File(label="download results")

    classify_btn.click(
        classify_and_log, 
        inputs=[image_input],
        outputs=[log_output]
    )


demo.launch()