Object detection from scratch in python

 Objective: Find out the number of objects from the image and                          give the distinct image to a distinct color.

Method and Experiment: First, we used imread() function to read the image. The next same image was converted into the grayscale by using cvtColor() function. Then used the np.unique() function that gives us distinct elements in sorted order. After that, we used cv2.threshold() function to assign 0 and 1. Then apply the algorithm to identify the object in the image. 

We used m-connected algorithm to identify the object from the image.

Note: we ran the code into google colab. If you prefer the same platform then you don’t need to change anything except the path of the image where the image is reading.

Code : 

Object Detection algorithm

        
img = cv2.imread('/content/cc.jpg')
img=cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) #convert gray scale
print(np.unique(img))   
img = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)[1]
cv2_imshow(img)

labels=random.sample(range(1,255 ), 254)


x=0
for i in range(160):
  for j in range(240):
    if(i==0 and j==0 and img[i][j]==255):
      img[i][j]=labels[x]
      x+=1
    else:
      if(i==0):
        if(img[i][j]==255):
          if(img[i][j-1]!=0 ):
            img[i][j]=img[i][j-1]
          else:
            img[i][j]=labels[x]
            x+=1
      if(j==0):
        if(img[i][j]==255):
          if(img[i-1][j]!=0):
            img[i][j]=img[i-1][j];
          else:
            img[i][j]=labels[x]
            x+=1
            
      else:
        if(img[i][j]==255):
          if(img[i-1][j]!=0):
            img[i][j]=img[i-1][j]
          elif(img[i][j-1]!=0):
            img[i][j]=img[i][j-1]
          else:
            img[i][j]=labels[x]
            x+=1




cv2_imshow(img)   


num_epochs = 15
for epoch in range(num_epochs):
  Dict={}
  for i in range(160):
    for j in range(240):
      if(img[i][j]!=0):
        if(img[i][j-1]!=0 and img[i-1][j]!=0):
          if(img[i][j-1]!=img[i-1][j]):
            Dict[img[i][j-1]]=img[i-1][j]

  # replacing multiple label to single label
  for key in Dict.keys():
    for i in range(160):
      for j in range(240):
        if(img[i][j] == Dict[key]):
          img[i][j] = key
cv2_imshow(img)
labels = np.unique(img)
print(labels)
img_index = copy.deepcopy(img)
for index, label in zip(range(len(labels)),labels):
  # print(index,label)
  for i in range(img.shape[0]):
    for j in range(img.shape[1]):
      if img_index[i][j] == label:
        img_index[i][j] = index

labels = np.unique(img_index)

labels=np.unique(img_index)
label_hue = np.uint8(179*img/np.max(labels))
blank_ch = 255*np.ones_like(label_hue)
labeled_img = cv2.merge([label_hue, blank_ch, blank_ch])
labeled_img = cv2.cvtColor(labeled_img, cv2.COLOR_HSV2BGR)
print(img.shape,labeled_img.shape)
# set bg label to black
labeled_img[label_hue==0] = 0

cv2_imshow(img)
# cv2_imshow(labeled_img)
# Showing Original Image
plt.imshow(cv2.cvtColor(labeled_img, cv2.COLOR_BGR2RGB))
plt.axis("off")
plt.title("labelled img")
plt.show()

        
        

 
    

Input image of the program :

Output Image of the program :