image_similarity.py

from sentence_transformers import SentenceTransformer, util

class ImageSimilarity(object):
    def __init__(self, minimum_commutative_image_diff):
        self.minimum_commutative_image_diff = minimum_commutative_image_diff    

    def check(self, pil_images):
        results = []

        # Load the OpenAI CLIP Model
        print('Loading CLIP Model...')
        model = SentenceTransformer('clip-ViT-B-32')

        print("Images:", len(pil_images))
        encoded_image = model.encode([image["pil"] for image in pil_images], batch_size=128, convert_to_tensor=True, show_progress_bar=True)

        # Now we run the clustering algorithm. This function compares images aganist 
        # all other images and returns a list with the pairs that have the highest 
        # cosine similarity score
        processed_images = util.paraphrase_mining_embeddings(encoded_image)
        NUM_SIMILAR_IMAGES = 10 

        # =================
        # DUPLICATES
        # =================
        # print('Finding duplicate images...')
        # Filter list for duplicates. Results are triplets (score, image_id1, image_id2) and is scorted in decreasing order
        # A duplicate image will have a score of 1.00
        # It may be 0.9999 due to lossy image compression (.jpg)
        # duplicates = [image for image in processed_images if image[0] >= 0.999]

        # Output the top X duplicate images
        # for score, image_id1, image_id2 in duplicates[0:NUM_SIMILAR_IMAGES]:
        #     print("\nScore: {:.3f}%".format(score * 100))
        #     print(pil_images[image_id1])
        #     print(pil_images[image_id2])

        # =================
        # NEAR DUPLICATES
        # =================
        print('Finding near duplicate images...')
        # Use a threshold parameter to identify two images as similar. By setting the threshold lower, 
        # you will get larger clusters which have less similar images in it. Threshold 0 - 1.00
        # A threshold of 1.00 means the two images are exactly the same. Since we are finding near 
        # duplicate images, we can set it at 0.99 or any number 0 < X < 1.00.
        threshold = 0.90
        near_duplicates = [image for image in processed_images if image[0] < threshold]

        for score, image_id1, image_id2 in near_duplicates[0:NUM_SIMILAR_IMAGES]:
            results.append({
                'score': score,
                'image1': pil_images[image_id1]["key"],
                'image2': pil_images[image_id2]["key"]
            })
            # print("\nScore: {:.3f}%".format(score * 100))
            # print(pil_images[image_id1]["key"])
            # print(pil_images[image_id2]["key"])
        
        return results