Frame extraction in Computer Vision - ML Experiment: Train & Evaluate

Experiment - Frame extraction

Problem:You want to extract frames from a video file to analyze or process them individually.

Current Metrics:The current method extracts frames but is very slow, taking about 10 seconds per frame on average.

Issue:The frame extraction process is inefficient and slow, making it hard to process videos quickly.

Your Task

Improve the frame extraction speed to extract frames at least 5 times faster without losing any frames.

You must use Python and OpenCV library.

You cannot reduce the number of frames extracted.

The output frames must be saved as image files.

Hint 1

Hint 2

Hint 3

Solution

Computer Vision

import cv2
import time

video_path = 'input_video.mp4'
cap = cv2.VideoCapture(video_path)

frame_count = 0
start_time = time.time()

while True:
    ret, frame = cap.read()
    if not ret:
        break
    cv2.imwrite(f'frame_{frame_count:04d}.jpg', frame)
    frame_count += 1

cap.release()
end_time = time.time()
print(f'Extracted {frame_count} frames in {end_time - start_time:.2f} seconds')

Used OpenCV's VideoCapture to read frames efficiently.

Removed any extra processing inside the loop.

Saved frames immediately after reading to avoid memory overhead.

Results Interpretation

Before: 1 frame took ~10 seconds to extract.
After: 300 frames extracted in 5.8 seconds (~0.02 seconds per frame).

Efficient use of video reading libraries and minimizing operations inside loops can greatly speed up frame extraction.

Bonus Experiment

Try extracting frames at a fixed interval, for example, every 10th frame, to reduce the number of frames saved.

💡 Hint

Use a frame counter and save frames only when the counter modulo 10 equals zero.