OpenCV 3 Computer Vision with Python Cookbook

How to do it...

You need to follow these steps:

1. Import the modules:

import cv2import numpy as np

2. Define a classify function, which gets frames from the video, transforms them into tensors, feeds them to the neural network, and selects five categories with the highest probability:

def classify(video_src, net, in_layer, out_layer,              mean_val, category_names, swap_channels=False):    cap = cv2.VideoCapture(video_src)    t = 0        while True:        status_cap, frame = cap.read()        if not status_cap:            break        if isinstance(mean_val, np.ndarray):            tensor = cv2.dnn.blobFromImage(frame, 1.0, (224, 224),                       1.0, False);            tensor -= mean_val        else:            tensor = cv2.dnn.blobFromImage(frame, 1.0, (224, 224),                                   mean_val, swap_channels);        net.setInput(tensor, in_layer); prob = net.forward(out_layer); ...