(Python)用Mask R-CNN检测空闲车位

  拆分问题

当我们想要通过机器学习解决一个复杂的问题时，第一步是将问题分解为简单任务的序列。 然后，使用拆分的方法，我们可以从我们的机器学习工具箱中使用不同的工具来解决每一个较小的任务。 通过将几个小的解决方案链接到一个流水线中，于是我们将拥有一个可以执行复杂操作的系统。

  第1步：检测图像中的停车位

  检测图像中的汽车

import os
import numpy as np
import cv2



    
import mrcnn.config
import mrcnn.utils
from mrcnn.model import MaskRCNN
from pathlib import Path


# Configuration that will be used by the Mask-RCNN library
class MaskRCNNConfig(mrcnn.config.Config):
    NAME = "coco_pretrained_model_config"
    IMAGES_PER_GPU = 1
    GPU_COUNT = 1
    NUM_CLASSES = 1 + 80  # COCO dataset has 80 classes + one background class
    DETECTION_MIN_CONFIDENCE = 0.6


# Filter a list of Mask R-CNN detection results to get only the detected cars / trucks
def get_car_boxes(boxes, class_ids):
    car_boxes = []

    for i, box in enumerate(boxes):
        # If the detected object isn't a car / truck, skip it
        if class_ids[i] in [3, 8, 6]:
            car_boxes.append(box)

    return np.array(car_boxes)


# Root directory of the project
ROOT_DIR = Path(".")

# Directory to save logs and trained model
MODEL_DIR = os.path.join(ROOT_DIR, "logs")

# Local path to trained weights file
COCO_MODEL_PATH = os.path.join(ROOT_DIR, "mask_rcnn_coco.h5")

# Download COCO trained weights from Releases if needed
if not os.path.exists(COCO_MODEL_PATH):
    mrcnn.utils.download_trained_weights(COCO_MODEL_PATH)




    
# Directory of images to run detection on
IMAGE_DIR = os.path.join(ROOT_DIR, "images")

# Video file or camera to process - set this to 0 to use your webcam instead of a video file
VIDEO_SOURCE = "test_images/parking.mp4"

# Create a Mask-RCNN model in inference mode
model = MaskRCNN(mode="inference", model_dir=MODEL_DIR, config=MaskRCNNConfig())

# Load pre-trained model
model.load_weights(COCO_MODEL_PATH, by_name=True)

# Location of parking spaces
parked_car_boxes = None

# Load the video file we want to run detection on
video_capture = cv2.VideoCapture(VIDEO_SOURCE)

# Loop over each frame of video
while video_capture.isOpened():
    success, frame = video_capture.read()
    if not success:
        break

    # Convert the image from BGR color (which OpenCV uses) to RGB color
    rgb_image = frame[:, :, ::-1]

    # Run the image through the Mask R-CNN model to get results.
    results = model.detect([rgb_image], verbose=0)

    # Mask R-CNN assumes we are running detection on multiple images.
    # We only passed in one image to detect, so only grab the first result.
    r = results[0]

    # The r variable will now have the results of detection:
    # - r['rois'] are the bounding box of each detected object
    # - r['class_ids'] are the class id (type) of each detected object
    # - r['scores'] are the confidence scores for each detection
    # - r['masks'] are the object masks for each detected object (which gives you the object outline)

    # Filter the results to only grab the car / truck bounding boxes
    car_boxes = get_car_boxes(r['rois'], r['class_ids'])

    print("Cars found in frame of video:")

    # Draw each box on the frame
    for box in car_boxes:
        print("Car: ", box)

        y1, x1, y2, x2 = box

        # Draw the box
        cv2.rectangle(frame, (x1, y1), (x2, y2), (0


    
, 255, 0), 1)

    # Show the frame of video on the screen
    cv2.imshow('Video', frame)

    # Hit 'q' to quit
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# Clean up everything when finished
video_capture.release()
cv2.destroyAllWindows()

Cars found in frame of video:
Car:  [492 871 551 961]
Car:  [450 819 509 913]
Car:  [411 774 470 856]

  检测空的停车位

   # Filter the results to only grab the car / truck bounding boxes
    car_boxes = get_car_boxes(r['rois'], r['class_ids'])

    # See how much cars overlap with the known parking spaces
    overlaps = mrcnn.utils.compute_overlaps(car_boxes, parking_areas)

    print(overlaps)

[
 [1.         0.07040032 0.         0.]
 [0.07040032 1.         0.07673165 0.]
 [0.         0.         0.02332112 0.]
]

  发送短信

pip3 install twilio

from twilio.rest import Client

# Twilio account details
twilio_account_sid = 'Your Twilio SID here'
twilio_auth_token = 'Your Twilio Auth Token here'
twilio_source_phone_number = 'Your Twilio phone number here'

# Create a Twilio client object instance
client = Client(twilio_account_sid, twilio_auth_token)




    
# Send an SMS
message = client.messages.create(
    body="This is my SMS message!",
    from_=twilio_source_phone_number,
    to="Destination phone number here"
)

import os
import numpy as np
import cv2
import mrcnn.config
import mrcnn.utils
from mrcnn.model import MaskRCNN
from pathlib import Path
from twilio.rest import Client


# Configuration that will be used by the Mask-RCNN library
class MaskRCNNConfig(mrcnn.config.Config):
    NAME = "coco_pretrained_model_config"
    IMAGES_PER_GPU = 1
    GPU_COUNT = 1
    NUM_CLASSES = 1 + 80  # COCO dataset has 80 classes + one background class
    DETECTION_MIN_CONFIDENCE = 0.6


# Filter a list of Mask R-CNN detection results to get only the detected cars / trucks
def get_car_boxes(boxes, class_ids):
    car_boxes = []

    for i, box in enumerate(boxes):
        # If the detected object isn't a car / truck, skip it
        if class_ids[i] in


    
 [3, 8, 6]:
            car_boxes.append(box)

    return np.array(car_boxes)


# Twilio config
twilio_account_sid = 'YOUR_TWILIO_SID'
twilio_auth_token = 'YOUR_TWILIO_AUTH_TOKEN'
twilio_phone_number = 'YOUR_TWILIO_SOURCE_PHONE_NUMBER'
destination_phone_number = 'THE_PHONE_NUMBER_TO_TEXT'
client = Client(twilio_account_sid, twilio_auth_token)


# Root directory of the project
ROOT_DIR = Path(".")

# Directory to save logs and trained model
MODEL_DIR = os.path.join(ROOT_DIR, "logs")

# Local path to trained weights file
COCO_MODEL_PATH = os.path.join(ROOT_DIR, "mask_rcnn_coco.h5")

# Download COCO trained weights from Releases if needed
if not os.path.exists(COCO_MODEL_PATH):
    mrcnn.utils.download_trained_weights(COCO_MODEL_PATH)

# Directory of images to run detection on
IMAGE_DIR = os.path.join(ROOT_DIR, "images")

# Video file or camera to process - set this to 0 to use your webcam instead of a video file
VIDEO_SOURCE = "test_images/parking.mp4"

# Create a Mask-RCNN model in inference mode
model = MaskRCNN(mode="inference", model_dir=MODEL_DIR, config=MaskRCNNConfig())

# Load pre-trained model
model.load_weights(COCO_MODEL_PATH, by_name=True)

# Location of parking spaces
parked_car_boxes = None

# Load the video file we want to run detection on
video_capture = cv2.VideoCapture(VIDEO_SOURCE)

# How many frames of video we've seen in a row with a parking space open
free_space_frames = 0

# Have we sent an SMS alert yet?
sms_sent = False

# Loop over each frame of video
while video_capture.isOpened():
    success, frame = video_capture.read()
    if not success:
        break

    


    
# Convert the image from BGR color (which OpenCV uses) to RGB color
    rgb_image = frame[:, :, ::-1]

    # Run the image through the Mask R-CNN model to get results.
    results = model.detect([rgb_image], verbose=0)

    # Mask R-CNN assumes we are running detection on multiple images.
    # We only passed in one image to detect, so only grab the first result.
    r = results[0]

    # The r variable will now have the results of detection:
    # - r['rois'] are the bounding box of each detected object
    # - r['class_ids'] are the class id (type) of each detected object
    # - r['scores'] are the confidence scores for each detection
    # - r['masks'] are the object masks for each detected object (which gives you the object outline)

    if parked_car_boxes is None:
        # This is the first frame of video - assume all the cars detected are in parking spaces.
        # Save the location of each car as a parking space box and go to the next frame of video.
        parked_car_boxes = get_car_boxes(r['rois'], r['class_ids'])
    else:
        # We already know where the parking spaces are. Check if any are currently unoccupied.

        # Get where cars are currently located in the frame
        car_boxes = get_car_boxes(r['rois'], r['class_ids'])

        # See how much those cars overlap with the known parking spaces
        overlaps = mrcnn.utils.compute_overlaps(parked_car_boxes, car_boxes)

        # Assume no spaces are free until we find one that is free
        free_space = False

        # Loop through each known parking space box
        for parking_area, overlap_areas in zip(parked_car_boxes, overlaps):

            # For this parking space, find the max amount it was covered by any
            # car that was detected in our image (doesn't really matter which car)
            max_IoU_overlap = np.max(overlap_areas)

            # Get the top-left and bottom-right coordinates of the parking area
            y1, x1, y2, x2 = parking_area

            # Check if the parking space is occupied by seeing if any car overlaps
            # it by more than 0.15 using IoU
            if max_IoU_overlap < 0.15


    
:
                # Parking space not occupied! Draw a green box around it
                cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 255, 0), 3)
                # Flag that we have seen at least one open space
                free_space = True
            else:
                # Parking space is still occupied - draw a red box around it
                cv2.rectangle(frame, (x1, y1), (x2, y2), (0, 0, 255), 1)

            # Write the IoU measurement inside the box
            font = cv2.FONT_HERSHEY_DUPLEX
            cv2.putText(frame, f"{max_IoU_overlap:0.2}", (x1 + 6, y2 - 6), font, 0.3, (255, 255, 255))

        # If at least one space was free, start counting frames
        # This is so we don't alert based on one frame of a spot being open.
        # This helps prevent the script triggered on one bad detection.
        if free_space:
            free_space_frames += 1
        else:
            # If no spots are free, reset the count
            free_space_frames = 0

        # If a space has been free for several frames, we are pretty sure it is really free!
        if free_space_frames > 10:
            # Write SPACE AVAILABLE!! at the top of the screen
            font = cv2.FONT_HERSHEY_DUPLEX
            cv2.putText(frame, f"SPACE AVAILABLE!", (10, 150), font, 3.0, (0


    
, 255, 0), 2, cv2.FILLED)

            # If we haven't sent an SMS yet, sent it!
            if not sms_sent:
                print("SENDING SMS!!!")
                message = client.messages.create(
                    body="Parking space open - go go go!",
                    from_=twilio_phone_number,
                    to=destination_phone_number
                )
                sms_sent = True

        # Show the frame of video on the screen
        cv2.imshow('Video', frame)

    # Hit 'q' to quit
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

# Clean up everything when finished
video_capture.release()
cv2.destroyAllWindows()