Deduplicate Images Based on Their Perceptual Hash

phash_dedup.py

import pathlib

import cv2 as cv
import fire
import imagehash
from PIL import Image


def dedup(video_uri: str, output_dir: str, min_hash_difference_between_key_frames: float=10):
    cap = cv.VideoCapture(video_uri)
    if not cap.isOpened():
        raise ValueError(f"Cannot open ${video_uri}")

    output_dir = pathlib.Path(output_dir)
    output_dir.mkdir(exist_ok=True)

    base_image_hash = None
    idx = 0
    while True:
        valid, frame = cap.read()
        if not valid:
            break
        image = cv.cvtColor(frame, cv.COLOR_BGR2RGB)
        image = Image.fromarray(image)
        image_hash = imagehash.phash(image)
        if base_image_hash is None or (
                image_hash - base_image_hash >= min_hash_difference_between_key_frames):
           base_image_hash = image_hash
           print(f"Identified frame {idx} to be a key frame.")
           image.save(output_dir/f"{idx:05}.jpg")
        idx += 1
    cap.release()

if __name__ == "__main__":
    fire.Fire(dedup)