What does cv2.dnn_Net.forward() function return exactly?

Question

I am using cv2.dnn_Net.forward() (forward propagation) for the face detecting pre-trained model. However, I couldn't grasp the idea of utilizing the returned variable from cv2.dnn_Net.forward() function (which is detections variable from below).

face_net = cv2.dnn.readNet(face_prototxt_path, face_weights_path) 
image = cv2.imread(args["image_path"])
(h, w) = image.shape[:2]
blob = cv2.dnn.blobFromImage(image, 1.0, (300, 300),
(104.0, 177.0, 123.0))

face_net.setInput(blob)   
detections = face_net.forward() # cv2.dnn_Net.forward() function

# Utilizing 'detections' variable
confidence = detections[0, 0, i, 2]
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])

What does cv2.dnn_Net.forward() return exactly and how can they be utilized as detections[0, 0, i, 2] and detections[0, 0, i, 3:7]?

Answer 1

cv2.dnn.readNet takes your weight file and configuaration file of your model to load your saved model.

net.forward() - Runs a forward pass to compute the net output.

your detection i.e net.forward() will give Numpy ndarray as output which you can use to plot the box on the given input image.