Can you please help me how to fix the bounding box based on the picture below using the flask code? Please provide the code. Thanks

Here is the code.

import cv$2$

import math

from deep$\_$sort import build$\_$tracker # Assuming DeepSORT is installed and this is the correct import

app $=$ Flask$(\_\_$name$\_\_)$

# model

model $=$ YOLO$("$best$.$pt$")$

# start webcam

cap $=$ cv$2.$VideoCapture$(0)$

cap.set$(3,640)$

cap.set$(4,480)$

# object classes

classNames $=["$person$"]$

# Initialize DeepSORT tracker

tracker $=$ build$\_$tracker$()$

def generate$\_$frames$()$:

frame$\_$count $=0$

fps $=0$

tick$1=$ cv$2.$getTickCount$()$

while True:

success, img $=$ cap.read$()$

img $=$ cv$2.$resize$($img$,(640,480))$

# Detect objects using YOLO

results $=$ model.track$($img$,$ stream$=$True, persist$=$True, tracker$=$"bytetrack.yaml"$)$

# Process detections for DeepSORT

boxes $=[]$

confidences $=[]$

class$\_$ids $=[]$

for r in results:

for box in r$.$boxes:

x$1,$ y$1,$ x$2,$ y$2=[$int$($x$)$ for x in box.xyxy$[0]]$

boxes.append$([$x$1,$ y$1,$ x$2-$x$1,$ y$2-$y$1])$ # Format: $[$x$,$ y$,$ w$,$ h$]$

confidences.append$($box$.$conf$[0])$

class$\_$ids.append$($int$($box$.$cls$[0]))$

# Update tracker

tracks $=$ tracker.update$($boxes$,$ confidences, img$)$

for track in tracks:

bbox $=$ track.to$\_$tlbr$()$ # Get the bounding box coordinates

cv$2.$rectangle$($img$,($int$($bbox$[0]),$ int$($bbox$[1])),($int$($bbox$[2]),$ int$($bbox$[3])),(0,255,0),5)$

cv$2.$putText$($img$,$ f$"\{$classNames$[$class$\_$ids$[$track$.$track$\_$id$]]\}$: $\{$track$.$track$\_$id$\}",$

$($int$($bbox$[0]),$ int$($bbox$[1])),$ cv$2.$FONT$\_$HERSHEY$\_$SIMPLEX, $1,(0,255,0),2)$

# Remaining code for FPS calculation and frame generation

# $...$

@app.route$(\text{'}/$video$\text{'})$

def video$()$:

return Response$($generate$\_$frames$(),$ mimetype$=$'multipart$/$x$-$mixed$-$replace; boundary$=$frame'$)$

@app.route$(\text{'}/\text{'})$

def index$()$:

return render$\_$template$(\text{'}$trial$.$html$\text{'})$

if name $==\text{'}\_\_$main$\_\_\text{'}$:

app.run$($host$=\text{'}0\mathrm{.}0\mathrm{.}0\mathrm{.}0\text{'},$ debug$=$True$)$

# before your while loop

tick$1=$ cv$2.$ getTickCount $()$

frame$\_$count $=\backslash$theta

fps $=0$

while True:

success, img$=$ cap.read $()$

img $=$cv$2.$resize$($img$,(640,480))$

#yolo detection

results $=$ model $($img$,$ stream$=$True$)$

#Deep sort process

boxes $=[]$

confidences $=[]$

class$\_$ids $=[]$

# coordinates

for r in results:

for box in boxes:

# bounding box

for x in box.xyxy$[0]]$ # convert to int values once

boxes.append $([$x$1,$y$1,$x$2-$x$1,$y$2-$y$1])$

# put box in cam

cv$2.$ rectangle$($img$,($x$1,$ y$1),($x$2,$ y$2),(0,255,0),5)$

# confidence

confidences $.$ append $($box $.$ conf$[\backslash$theta $])$

print$("$Confidence $--->",$ confidences$)$

# class id

class$\_$ids.append$($int$($box$.$cls$[0]))$

# Update tracker

tracks $=$ tracker. update$($boxes$,$ confidences, img$)$

for track in tracks:

Speed: $1\mathrm{.}0$ms preprocess, $69\mathrm{.}0$ms inference, $1\mathrm{.}0$ms postprocess per image at shape $(1,3,256,320)$

$0$: $256\backslash$times $3201$ person, $65\mathrm{.}0$ms

Speed: $1\mathrm{.}0$ms preprocess, $65\mathrm{.}0$ms inference, $1.$ Oms postprocess per image at shape $(1,3,256,320)$

$0$: $256\backslash$times $3201$ person, $70\mathrm{.}0$ms

Speed: $1\mathrm{.}0$ms preprocess, $70\mathrm{.}0$ms inference, $1.$ Oms postprocess per image at shape $(1,3,256,320)$

$0$: $256\backslash$times $3201$ person, $62\mathrm{.}0$ms

Speed: $1\mathrm{.}0$ms preprocess, $62\mathrm{.}0$ms inference, $1.$ Oms postprocess per image at shape $(1,3,256,320)$# before your while loop

tick$1=$ cv$2.$ getTickCount $()$

frame$\_$count $=$

fps $=0$

while True:

success, img$=$ cap.read $()$

img $=cv2.$resize$(img,(640,480))$

#yolo detection

results $=$ model $($img$,$ stream$=$True$)$

#Deep sort process

boxes $=[]$

confidences $=[]$

class$\_$ids $=[]$

# coordinates

for $r$ in results:

for box in boxes:

# bounding box

for $x$ in box.xyxy$[0]]$ # convert to int values once

boxes.append $([x1,y1,x2-x1,y2-y1])$

# put box in cam

cv$2.$ rectangle$($img$,($x$1,$ y$1),($x$2,$ y$2),(0,255,0),5)$

# confidence

confidences $.$ append $($box $.$ conf$[])$

print$("$Confidence $--->",$ confidences$)$

# class id

class$\_$ids.append$($int$($box$.$cls$[0]))$

# Update tracker

tracks $=$ tracker. update$($boxes$,$ confidences, img$)$

for track in tracks:

Speed: $1\mathrm{.}0$ms preprocess, $69\mathrm{.}0$ms inference, $1\mathrm{.}0$ms postprocess per image at shape $(1,3,256,320)$

$0$: $2563201$ person, $65\mathrm{.}0ms$

Speed: $1\mathrm{.}0ms$ preprocess, $65\mathrm{.}0ms$ inference, $1.$ Oms postprocess per image at shape $(1,3,256,320)$

$0$: $2563201$ person, $70\mathrm{.}0ms$

Speed: $1\mathrm{.}0$ms preprocess, $70\mathrm{.}0ms$ inference, $1.$ Oms postprocess per image at shape $(1,3,256,320)$

$0$: $2563201$ person,