Add cv2.line

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

I have a project for counting people in a shop, I saw some example but I would want to try to do by myself and understand what I do.
I konw to script in php/mysql/jquery/… but I just start with python and AI.

I have install jetson-inference.

I have created a new directory “example” and I run it with
cd jetson-inference
docker/run.sh --volume ~/example:/example
cd /example/
python3 example.py

In example.py I have add this script :

Blockquote
#!/usr/bin/python3
import numpy as np
import cv2
import jetson.inference
import jetson.utils

setup the network we are using

net = jetson.inference.detectNet(“ssd-mobilenet-v2”, threshold=0.5)

cap = cv2.VideoStream(src=1).start()

cap = cv2.VideoCapture(1)

cap = cv2.VideoCapture(‘video.mp4’)

cap.set(3,640)
cap.set(4,480)

while (True):
ret, frame = cap.read()
frame = cv2.resize(frame, (640, 480))
w = frame.shape[1]
h = frame.shape[0]
# to RGBA
# to float 32
input_image = cv2.cvtColor(frame, cv2.COLOR_RGB2RGBA).astype(np.float32)
# move the image to CUDA:
input_image = jetson.utils.cudaFromNumpy(input_image)

detections = net.Detect(input_image, w, h)
count = len(detections)
print("detected {:d} objects in image".format(len(detections)))

for detection in detections:
    print(detection)

# print out timing info
net.PrintProfilerTimes()
# Display the resulting frame
numpyImg = jetson.utils.cudaToNumpy(input_image, w, h, 4)
# now back to unit8
result = numpyImg.astype(np.uint8)
# Display fps
fps = 1000.0 / net.GetNetworkTime()
font = cv2.FONT_HERSHEY_SIMPLEX
line = cv2.LINE_AA
cv2.putText(result, "FPS: " + str(int(fps)) + ' | Detecting', (11, 20), font, 0.5, (32, 32, 32), 4, line)
cv2.putText(result, "FPS: " + str(int(fps)) + ' | Detecting', (10, 20), font, 0.5, (240, 240, 240), 1, line)
cv2.putText(result, "Total: " + str(count), (11, 45), font, 0.5, (32, 32, 32), 4, line)
cv2.putText(result, "Total: " + str(count), (10, 45), font, 0.5, (240, 240, 240), 1, line)
# show frames
cv2.imshow('frame', result)
if cv2.waitKey(1) & 0xFF == ord('q'):
    break

When everything done, release the capture

cap.release()
cv2.destroyAllWindows()

Blockquote

And now I want to add cv2.line to split in two spaces my frame.

I have try to put it out of the while and inside but nothing appen.
with this :
cv2.line(frame, (0, h // 2), (w, h // 2), (0, 255, 255), 2)

Could you help me for that, thanks in advance

2

Hi,
Please refer to below links related custom plugin implementation and sample:
https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/sampleOnnxMnistCoordConvAC

Thanks!

3

Hi @thierry_sco,

Please make sure you have given start point and end point dimensions in “cv2.line” correctly. And I believe “cv2.line” should be inside the while loop as “frame” data will be overwritten with new frame data per each iteration when you read video file.

Thank you.

4

Hi Guy,

Many thanks for your help, I never hope to receive so quickly some answer.

The solution was to not use frame in cv2.line but result.
cv2.line(result…

1 Like