The python version is 3.10.6 by default on ubuntu 22.04. It may have some compatibility issues with OpenVINO. In this article, I will demonstrate how to install Openvino (CPU only) on a fresh ubuntu 22.04. The CPU model I used in this demonstration is the Intel i5-8250u.
Step 1: Add this repository to your ubuntu system
sudo add-apt-repository ppa:deadsnakes/ppa
Step 2: Update the list of available packages and install python 3.9
sudo apt update
sudo apt install python3.9-dev -y
sudo apt install python3.9-venv -y
sudo apt install python3-pip -y
Step 3: Create a virtual environment for OpenVINO
cd ~
python3.9 -m venv openvino_env
Step 4: Activate the virtual environment and install the required packages using pip
source ~/openvino_env/bin/activate
python3.9 -m pip install --upgrade pip
python3.9 -m pip install openvino-dev[tensorflow2,pytorch,caffe]
Step 5: Install git and clone the MobileNet-SSD project from GitHub for testing
cd ~
git clone https://github.com/chuanqi305/MobileNet-SSD.git
Step 6: Move to the MobileNet-SSD folder and create a script for testing. The script I used is based on an article written by Adrian Rosebrock on Pyimagesearch in 2019. I made some modifications to it.
cd MobileNet-SSD
sudo nano detection_image.py
# import the necessary packages
import numpy as np
import argparse
import time
import cv2
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-p", "--prototxt", required=True,
help="path to Caffe 'deploy' prototxt file")
ap.add_argument("-m", "--model", required=True,
help="path to Caffe pre-trained model")
ap.add_argument("-c", "--confidence", type=float, default=0.2,
help="minimum probability to filter weak detections")
ap.add_argument("-u", "--movidius", type=bool, default=0,
help="boolean indicating if the Movidius should be used")
args = vars(ap.parse_args())
# initialize the list of class labels MobileNet SSD was trained to
# detect, then generate a set of bounding box colors for each class
CLASSES = ["background", "aeroplane", "bicycle", "bird", "boat",
"bottle", "bus", "car", "cat", "chair", "cow", "diningtable",
"dog", "horse", "motorbike", "person", "pottedplant", "sheep",
"sofa", "train", "tvmonitor"]
COLORS = np.random.uniform(0, 255, size=(len(CLASSES), 3))
# load our serialized model from disk
print("[INFO] loading model...")
net = cv2.dnn.readNetFromCaffe(args["prototxt"], args["model"])
#CPU only
net.setPreferableTarget(cv2.dnn.DNN_TARGET_CPU)
image_path = "./images/000001.jpg"
frame = cv2.imread(image_path)
time.sleep(2.0)
# grab the frame dimensions and convert it to a blob with a maximum width of 300 pixels and a maximum height of 300 pixels
h, w, _ = frame.shape
frame = cv2.resize(frame, (0,0), fx=300/w, fy=300/h)
blob = cv2.dnn.blobFromImage(frame, 0.007843, (300, 300), 127.5)
# pass the blob through the network and obtain the detections and
# predictions
net.setInput(blob)
detections = net.forward()
# loop over the detections
for i in np.arange(0, detections.shape[2]):
# extract the confidence (i.e., probability) associated with
# the prediction
confidence = detections[0, 0, i, 2]
# filter out weak detections by ensuring the `confidence` is
# greater than the minimum confidence
if confidence > args["confidence"]:
# extract the index of the class label from the
# `detections`, then compute the (x, y)-coordinates of
# the bounding box for the object
idx = int(detections[0, 0, i, 1])
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
# draw the prediction on the frame
label = "{}: {:.2f}%".format(CLASSES[idx],
confidence * 100)
cv2.rectangle(frame, (startX, startY), (endX, endY),
COLORS[idx], 2)
y = startY - 15 if startY - 15 > 15 else startY + 15
cv2.putText(frame, label, (startX, y),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, COLORS[idx], 2)
# show the output frame
cv2.imshow("Frame", frame)
#halt the program for 5s before exit
key = cv2.waitKey(5000)
# do a bit of cleanup
cv2.destroyAllWindows()
Press ctrl + X to save the file Step 8: Test the program with this command
python3.9 detection_image.py --prototxt deploy.prototxt --model mobilenet_iter_73000.caffemodel
Optional: Try this code to use the Caffe module with a web camera or video. It is based on an article by Adrian Rosebrock on Pyimagesearch in 2019, and I modified it. It is not perfect, as the video path is hard-coded; you may change the program by yourself to use a dynamic path.
sudo nano detection.py
# import the necessary packages
import numpy as np
import argparse
import time
import cv2
# construct the argument parse and parse the arguments
ap = argparse.ArgumentParser()
ap.add_argument("-p", "--prototxt", required=True,
help="path to Caffe 'deploy' prototxt file")
ap.add_argument("-m", "--model", required=True,
help="path to Caffe pre-trained model")
ap.add_argument("-c", "--confidence", type=float, default=0.2,
help="minimum probability to filter weak detections")
ap.add_argument("-u", "--movidius", type=bool, default=0,
help="boolean indicating if the Movidius should be used")
args = vars(ap.parse_args())
# initialize the list of class labels MobileNet SSD was trained to
# detect, then generate a set of bounding box colors for each class
CLASSES = ["background", "aeroplane", "bicycle", "bird", "boat",
"bottle", "bus", "car", "cat", "chair", "cow", "diningtable",
"dog", "horse", "motorbike", "person", "pottedplant", "sheep",
"sofa", "train", "tvmonitor"]
COLORS = np.random.uniform(0, 255, size=(len(CLASSES), 3))
# load our serialized model from disk
print("[INFO] loading model...")
net = cv2.dnn.readNetFromCaffe(args["prototxt"], args["model"])
#CPU only
net.setPreferableTarget(cv2.dnn.DNN_TARGET_CPU)
# initialize the video stream, allow the cammera sensor to warmup
print("[INFO] starting video stream...")
#change this to your web camera streaming path
url = "rtsp://192.168.1.202:554/"
vs = cv2.VideoCapture(url)
#for video
#video_path = "./your_video_path.avi"
#vs = cv2.VideoCapture(video_path)
time.sleep(2.0)
# loop over the frames from the video stream
while True:
# grab the frame from the threaded video stream and resize it
# to have a maximum width of 300 pixels and a maximum height of 300 pixels
ret, frame = vs.read()
# grab the frame dimensions and convert it to a blob
h, w, _ = frame.shape
frame = cv2.resize(frame, (0,0), fx=300/w, fy=300/h)
blob = cv2.dnn.blobFromImage(frame, 0.007843, (300, 300), 127.5)
# pass the blob through the network and obtain the detections and
# predictions
net.setInput(blob)
detections = net.forward()
# loop over the detections
for i in np.arange(0, detections.shape[2]):
# extract the confidence (i.e., probability) associated with
# the prediction
confidence = detections[0, 0, i, 2]
# filter out weak detections by ensuring the `confidence` is
# greater than the minimum confidence
if confidence > args["confidence"]:
# extract the index of the class label from the
# `detections`, then compute the (x, y)-coordinates of
# the bounding box for the object
idx = int(detections[0, 0, i, 1])
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
# draw the prediction on the frame
label = "{}: {:.2f}%".format(CLASSES[idx],
confidence * 100)
cv2.rectangle(frame, (startX, startY), (endX, endY),
COLORS[idx], 2)
y = startY - 15 if startY - 15 > 15 else startY + 15
cv2.putText(frame, label, (startX, y),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, COLORS[idx], 2)
# show the output frame
cv2.imshow("Frame", frame)
key = cv2.waitKey(1) & 0xFF
# if the `q` key was pressed, break from the loop
if key == ord("q"):
break
# do a bit of cleanup
cv2.destroyAllWindows()
vs.stop()
Press ctrl + X to save the file Test the program with this command
python3.9 detection.py --prototxt deploy.prototxt --model mobilenet_iter_73000.caffemodel
Congratulation if your OpenVINO works appropriately with the Caffe module. I have not tested other modules yet, such as PyTorch and TensorFlow2. You may try to use these modules by yourself.
Leave a comment below if you have any trouble. I am willing to help you if I have time.