I$\text{'}$m currently learning how to do custom dataset object detection by using Yolov$5.$ This is my current code and I am using PyCharm. It does not show any bounding box or identification when I run an mp$4$ video. But when I use the terminal by using python detect.py $--$weights best.pt $--$conf $0\mathrm{.}45--$img$-$size $420--$source CustomData$/$tridaxnew$.$mp$4...$ it brings me the results I want. However, if I do it in the python script, it just shows the video of it but it does not show any bounding box and identification. I did this so I can check if the yoloV$5$ detection works in real$-$time when i use my webcam.

This is my code in main.py$.$ Please show me how to fix it and explain the changes in the code.

import torch

from matplotlib import pyplot as plt

import numpy as np

import cv$2$

model $=$ torch.hub.load$(\text{'}$ultralytics$/$yolov$5\text{'},$ 'custom', path$=$'yolov$5/$best$.$pt$\text{'},$force$\_$reload$=$True$)$

#real time detections

cap $=$ cv$2.$VideoCapture$(\text{'}$tridaxnew$.$mp$4\text{'})$ #This video should have bounding box and identification but it is not

while cap.isOpened$()$:

ret, frame $=$ cap.read$()$

#make detections

results $=$ model$($frame$)$

cv$2.$imshow$(\text{'}$ObjectDetection$\text{'},$ np$.$squeeze$($results$.$render$()))$

if cv$2.$waitKey$(10)$ & $0$xFF $==$ ord$(\text{'}$q$\text{'})$:

break

cap.release$()$

cv$2.$destroyAllWindows$()$