As you know that YOLO(You Only Look Once) is real time object detection algorithm.

You can implement this by own just follow this.

Step1: Dependencies.

1.python 3.6 (downloadchose 3.6.0).
2.OpenCV(downloas).
3.tensorflow (If you have gpu then use gpu version)

Step2: Clone and Download this repository,Or you can download it from here.

Step3: Build the cython extensions.

Open cmd at darkflow-master folder and type

python setup.py build_ext --inplace

Step4: Download the weights.

1. Downloads the weight according to your model here we will use YOLOv2 so weights(Download YOLOv2 608x608 weight).

2. Create a folder in darkflow-master folder and name it bin.

3. Paste the downloaded weight in this folder(or you can direct download in this folder😁).

Step5: Nothing just start it.

Here the most interesting part comes...\

If you want your model to render a video then copy the video and paste it under the darkflow-master folder. and use this command

python flow --model cfg/yolo.cfg --load bin/yolov2.weights --demo videofile.mp4 --gpu 1.0 --saveVideo

if you get some error like ModuleNotFoundError: No module named 'nms' then download the repo from here and run build_ext properly.

Code Description.

python flow --model cfg/yolo.cfg :- Which model you want to use give it path to the model.

--load bin/yolov2.weight :- Path of the weight check the downloaded weight name if this is yolo.weight then make yolov2.weight to yolo.weight is the above code.

--demo videofile.mp4 :- video file name name.

--gpu 1.0 :- tensorflow gpu if you have tensorflow gpu then give it leave it.

--saveVideo :- This is for save the video file, your rended video file will be video.avi named.

Now if you want your web camera to detect object then run this command

python flow --model cfg/yolo.cfg --load bin/yolov2.weights --demo camera --gpu 1.0 --saveVideo

just use camera keyword.

If you want to detect object from a image then go for it,

Open you jupyter notebook and start typing...

from darkflow.net.build import TFNet
import cv2

options = {"model": "cfg/yolo.cfg", "load": "bin/yolov2.weights", "threshold": 0.3}

tfnet = TFNet(options)

imgcv = cv2.imread("./sample_img/dog.jpg")
result = tfnet.return_predict(imgcv)
print(result)

Their is no code for run yolov2 on single image from cmd (or idk) so i try this just run yolov2photo.py ,before that open the yolov2photo.py file and modify the options according to your cfg,weight location and if you have gpu the make "gpu":1.0.

import cv2
import matplotlib.pyplot as plt
from darkflow.net.build import TFNet
import argparse

# the cfg file and weights location change this thing according to your model location
model = {"model": "cfg/yolov2.cfg",
           "load": "yolov2.weights",
           "threshold": 0.4,
         "gpu":0}

# creating the object
tfnet = TFNet(model)

# to get the image path
parser = argparse.ArgumentParser()
parser.add_argument('--img',type=str,help='path of the image')
arg = parser.parse_args()


imgc = cv2.imread(arg.img) # read the image
imgcv = cv2.cvtColor(imgc,cv2.COLOR_RGB2BGR)
result = tfnet.return_predict(imgcv) # predict the classes and cordinates of the oject

# This is for draw the bounding box around the predicted classes 
tl = []
br = []
labels = []
confidence = []
cf_cor = []

for i in range(len(result)):
    topleft = (result[i]['topleft']['x'],result[i]['topleft']['y']) # to get the topleft cordinates in a tuple
    bottomright = (result[i]['bottomright']['x'],result[i]['bottomright']['y']) # to get the bottomright cordinates in a tuple
    label = (result[i]['label']) # to get the labels from the predicted class ,it's in the form of dictionary
    conf = str(round(result[i]['confidence'],2))
    
    # to add the confidence score in middle
    st = result[i]['topleft']['x'] 
    nd = result[i]['bottomright']['x']
    mid_x = (nd-st)//2 + st # mid point of the top box line
    mid_y = result[i]['topleft']['y']
    
    cordi = (mid_x,mid_y)
    cf_cor.append(cordi)                 
    tl.append(topleft) # append the tuples in the list
    br.append(bottomright)
    labels.append(label)
    confidence.append(conf)
    img2 = cv2.rectangle(imgcv,tl[i],br[i],(0,255,0),3) # draw rectangles around the classes here we pass image,topleft cordinates ,bottomright cordinates ,which colour box we want and how thik the line
    img2 = cv2.putText(imgcv,labels[i],tl[i],cv2.FONT_HERSHEY_COMPLEX,1, (255 ,0 ,0), 2) # putting the label on the topleft corner
    img2 = cv2.putText(imgcv,confidence[i],cf_cor[i],cv2.FONT_ITALIC,1, (255 ,0 ,0), 2) # putting the confidence score
    

    
img3 = cv2.cvtColor(img2,cv2.COLOR_BGR2RGB)  # convert the image in RGB format
cv2.imshow('prediction',img3)
cv2.waitKey(0) # waitkey for hold the image in display until user press any key

if you do this thing on jupyter then open yolov2_show_img.ipynb do enjoy. It will predict the object with confident score in that given image.
Now enjoy ✌.