File size: 2,987 Bytes
ac1c6ae
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
from colordescriptor import ColorDescriptor
from CLIP import CLIPImageEncoder
import gradio as gr
import cv2
import numpy as np
from datasets import *

dataset = load_dataset("huggan/CelebA-faces")
candidate_subset = dataset["train"].select(range(1000)) # This is a small CBIR app! :D

def emb_dataset(dataset):
    # This function might need to be split up, to reduce start-up time of app
    # It could also use batches to increase speed
    # If indexes are saved in files, this is all not really necessary

    ## Color Embeddings
    cd = ColorDescriptor((8, 12, 3))
    dataset_with_embeddings = dataset.map(lambda row: {'color_embeddings': cd.describe(row["image"])}) # we assume that dataset has a column 'image'
    dataset_with_embeddings.add_faiss_index(column='color_embeddings')

    ## CLIP Embeddings
    clip_model = CLIPImageEncoder()
    dataset_with_embeddings = dataset.map(lambda row: {'clip_embeddings': clip_model.encode_image(row["image"])})
    dataset_with_embeddings.add_faiss_index(column='clip_embeddings')
    dataset_with_embeddings # Just to check, if okay

    return dataset_with_embeddings

dataset_with_embeddings = emb_dataset(candidate_subset)

# Main function, to find similar images
# TODO: allow different descriptor/embedding functions
# TODO: implement different distance measures

def get_neighbors(query_image, selected_descriptor, top_k=5):
    """Returns the top k nearest examples to the query image.

    Args:
        query_image: A PIL object representing the query image.
        top_k: An integer representing the number of nearest examples to return.

    Returns:
        A list of the top_k most similar images as PIL objects.
    """
    if  "Color Descriptor" in selected_descriptor:
        cd = ColorDescriptor((8, 12, 3))
        qi_embedding = cd.describe(query_image)
        qi_np = np.array(qi_embedding)
        scores, retrieved_examples = dataset_with_embeddings.get_nearest_examples(
            'color_embeddings', qi_np, k=top_k)
        images = retrieved_examples['image'] #retrieved images is a dict, with images and embeddings
        return images
    if "CLIP" in selected_descriptor:
        clip_model = CLIPImageEncoder()
        qi_embedding = clip_model.encode_image(query_image)
        scores, retrieved_examples = dataset_with_embeddings.get_nearest_examples(
            'clip_embeddings', qi_embedding, k=top_k)
        images = retrieved_examples['image']
        return images
    else:
        print("This descriptor is not yet supported :(")
        return []


# Define the Gradio Interface


iface = gr.Interface(
    fn=get_neighbors,
    inputs=[
        gr.Image(type="pil", label="Your Image"),
        gr.CheckboxGroup(["Color Descriptor", "LBP", "CLIP"], label="Descriptor method?"),
    ],
    outputs=gr.Gallery(),
    title="Image Similarity Gallery",
    description="Upload an image and get similar images",
    allow_flagging="never"
)

# Launch the Gradio interface
iface.launch()