upload files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+w2c11_j.npy filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,118 @@
+import gradio as gr
+import os
+import shutil
+import cv2
+import numpy as np
+
+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 = 512
+    width = img.shape[1]
+    height = img.shape[0]
+    if width > 1024 or height > 1024:
+        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)
+
+    w2c = np.load('w2c11_j.npy').astype(np.float16)
+    _, _, name_idx_img, _ = im2c(img, w2c)
+
+    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_colors = [COLOR_NAME[i] for i in top_3_values]
+    # print("Top 3 colors:", top_3_colors)
+    return top_3_values, top_3_colors
+
+
+def classify_and_log(images):
+    output_folder = "classified_results"
+    os.makedirs(output_folder, exist_ok=True)
+
+    category_folders = {i: os.path.join(output_folder, 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_folder, "log.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, category = get_top_names(img_array)
+            # print(category_folders[cat_id[0]], filename)
+            target_path = os.path.join(category_folders[cat_id[0]], 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} -> {category}\n")
+
+            results[cat_id[0]].append(target_path)
+
+    # output_images = [results[i] for i in range(11)]
+    # print(output_images)
+
+    return log_file
+
+
+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")
+
+    with gr.Row():
+        with gr.Column():
+            image_input = gr.File(
+                        label="Drag (Select) more than one photos",
+                        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 logs")
+
+    classify_btn.click(
+        classify_and_log, 
+        inputs=[image_input],
+        outputs=[log_output]
+    )
+
+
+demo.launch()

naming.py

@@ -0,0 +1,79 @@
+import numpy as np
+import cv2
+from collections import Counter
+
+
+COLOR_NAME = ['black', 'brown', 'blue', 'gray', 'green', 'orange', 'pink', 'purple', 'red', 'white', 'yellow']
+
+def im2c(im, w2c):
+    """
+    Convert an image to color name representation using a color-name matrix.
+    
+    Returns:
+        numpy.ndarray: Processed image based on color parameter.
+    """
+    # Define color name mappings
+    color_values = np.array([[  0,   0,   0], 
+                                [165,  81,  43], 
+                                [  0,   0, 255], 
+                                [127, 127, 127],
+                                [  0, 255,   0], 
+                                [255, 127,   0], 
+                                [255, 165, 216], 
+                                [191,   0, 191],
+                                [255,   0,   0], 
+                                [255, 255, 255], 
+                                [255, 255,   0]], dtype=np.uint8)
+    
+    
+    # Extract RGB channels
+    # RR, GG, BB = im[:, :, 0].flatten(), im[:, :, 1].flatten(), im[:, :, 2].flatten()
+    
+    # Compute index for w2c lookup
+    index_im = ((im[:, :, 0].flatten() // 8) + 32 * (im[:, :, 1].flatten()// 8) + 32 * 32 * (im[:, :, 2].flatten() // 8)).astype(np.int32)
+    
+    # w2cM = np.argmax(w2c, axis=1)
+    # name_idx_img = w2cM[index_im].reshape(im.shape[:2])
+    
+    # max_prob = w2c[np.arange(w2c.shape[0]), w2cM]
+    # max_prob_map = max_prob[index_im].reshape(im.shape[:2])
+    
+    prob_map = w2c[index_im, :].reshape((im.shape[0], im.shape[1], w2c.shape[1]))
+    max_prob_map = np.max(prob_map, axis=2)
+    name_idx_img = np.argmax(prob_map, axis=2)
+    
+    color_img = np.zeros_like(im).astype(np.uint8)
+        
+    for jj in range(im.shape[0]):
+        for ii in range(im.shape[1]):
+            color_img[jj, ii, :] = np.array(color_values[name_idx_img[jj, ii]])
+
+    return prob_map, max_prob_map, name_idx_img, color_img
+    
+
+if __name__ == "__main__":
+    # Load w2c matrix
+    w2c = np.load('w2c11_j.npy').astype(np.float16)  # Assuming 'w2c.mat' was converted to 'w2c.npy'
+
+
+    image_path = './test.jpg'
+    img = cv2.imread(image_path).astype(np.float32)
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+
+    prob_map, max_prob_img, name_idx_img, color_img = im2c(img, w2c)
+
+
+    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_colors = [COLOR_NAME[i] for i in top_3_values]
+
+    print("Top 3 colors:", top_3_colors)
+
+    cv2.imwrite('colormap_j.jpg', cv2.cvtColor(color_img, cv2.COLOR_BGR2RGB))
+
+    # print(prob_map.sum(axis=2))
+    # print(name_idx_img)
+    # print(max_prob_img.shape)
+    # print(color_img)
+

requirements.txt

@@ -0,0 +1,3 @@
+numpy
+cv2
+opencv-python

w2c11_j.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5ede48fed749d552edf1fd5d058266c713f20af3d4a484310103be91766856
+size 2883712