Python模块导入报错：视频人脸提取脚本调用失效排查

Let's work through your issues one by one and get your face extraction app up and running properly:

1. Fixing the TypeError: 'module' object is not callable #

The first error happens because you're trying to call the entire extractor module like a function (extractor()) in app.py. Modules aren't callable—you need to call specific functions inside the module instead.

2. Why your core logic isn't executing when imported #

Your frame extraction and face detection code is wrapped inside the if __name__ == "__main__" block in extractor.py. This block only runs when you run extractor.py directly (e.g., python extractor.py), not when you import it as a module into app.py. We'll fix this by moving the core workflow into a reusable function.

3. Hidden bugs blocking your functionality #

There are a few other issues in extractor.py that are stopping the core features from working:

• Path mismatch: Your extract_frames function saves frames to the frames/ folder, but your face detection functions are looking for frames in data/—so they can't find any files to process.
• Multiprocessing argument error: When passing args to multiprocessing.Process, you need to pass a tuple. args=(currentFrame_extract) (a list) will cause a parameter mismatch error for your find_faces_* functions.
• Redundant code: You have four nearly identical find_faces_* functions—we can simplify this into a single function that takes a starting frame as a parameter.
• Useless format call: The .format(i) in your save line is unnecessary (you're already using an f-string with currentFrame, and i isn't even used).

Fixed Code #

extractor.py

import cv2
import os
import face_recognition
from PIL import Image
import multiprocessing

# Create directories once at module load
for dir_name in ['frames', 'faces']:
    try:
        if not os.path.exists(dir_name):
            os.makedirs(dir_name)
            print(f"Created directory: {dir_name}")
    except OSError as e:
        print(f'Error creating directory {dir_name}: {e}')

def extract_frames(video_file_path):
    currentFrame_extract = 1
    video_capture = cv2.VideoCapture(video_file_path)
    if not video_capture.isOpened():
        print(f"Error: Could not open video file {video_file_path}")
        return 0

    while True:
        ret, frame = video_capture.read()
        if not ret:
            break
        name = f'frames/frame_{currentFrame_extract}.jpg'
        print(f"Extracting Frame {currentFrame_extract}, saving as {name}")
        cv2.imwrite(name, frame)
        currentFrame_extract += 1
    
    video_capture.release()
    cv2.destroyAllWindows()
    return currentFrame_extract - 1  # Return total frames extracted

def process_frames(start_frame, total_frames):
    currentFrame = start_frame
    while currentFrame <= total_frames:
        frame_path = f"frames/frame_{currentFrame}.jpg"
        if not os.path.exists(frame_path):
            print(f"Warning: Frame {currentFrame} not found, skipping")
            currentFrame += 4
            continue
            
        image = face_recognition.load_image_file(frame_path)
        face_locations = face_recognition.face_locations(image)
        
        if face_locations:
            top, right, bottom, left = face_locations[0]
            face_image = image[top:bottom, left:right]
            pil_image = Image.fromarray(face_image)
            face_save_path = f"faces/face_{currentFrame}.jpg"
            pil_image.save(face_save_path)
            print(f"Found face at Frame {currentFrame}, saved to {face_save_path}")
        
        currentFrame += 4

def run_face_extraction(video_file_path):
    total_frames = extract_frames(video_file_path)
    if total_frames == 0:
        print("No frames extracted, exiting")
        return

    # Create processes with correct tuple arguments
    processes = []
    for start in range(1, 5):
        p = multiprocessing.Process(target=process_frames, args=(start, total_frames))
        processes.append(p)
        p.start()

    # Wait for all processes to finish
    for p in processes:
        p.join()

    print("Frame extraction and face detection finished successfully.")

# Keep this for direct execution
if __name__ == "__main__":
    run_face_extraction("Video_3.mp4")

app.py

import extractor

# Call the exposed function from the module
extractor.run_face_extraction("Video_3.mp4")  # Replace with your video path if needed

Key Improvements #

• Reusable function: run_face_extraction is now the public entry point that can be called from app.py.
• Fixed paths: All frame references now use the frames/ directory.
• Simplified multiprocessing: Replaced four duplicate functions with a single process_frames function that takes a starting frame parameter.
• Error handling: Added checks for video file opening and frame existence to catch issues early.
• Cleaner directory setup: Combined directory creation into a loop to reduce redundancy.

内容的提问来源于stack exchange，提问作者marcr