3

I'm using tensorflow for object detection in the webcam and further I have to identify the coordinates of each of the identified object in the image. When I print the bounding boxes i.e. "boxes:" what i see a array of array and if I pull the first array within boxes array it gives [ymin,xmin,ymax,xmax] which is the coordinates of first object what i guess.

My Question: if there are three objects identified: Person, Chari, Backpack then from the boxes array how to get the coordinates for each of the identified objects

# Each box represents a part of the image where a particular object was detected.
    detection_boxes = detection_graph.get_tensor_by_name('detection_boxes:0')

# Actual detection.
      (boxes, scores, classes, num) = sess.run(
          [detection_boxes, detection_scores, detection_classes, num_detections],
          feed_dict={image_tensor: image_np_expanded})

Coordinates of the first Object

      ymin = int((boxes[0][0][0]*height))
      xmin = int((boxes[0][0][1]*width))
      ymax = int((boxes[0][0][2]*height))
      xmax = int((boxes[0][0][3]*width))

Boxes Array

[[[3.98103416e-01 2.58287102e-01 9.33272898e-01 7.91557074e-01]
  [5.12995601e-01 2.65186965e-01 8.48852396e-01 4.13343787e-01]
  [3.66972327e-01 7.32781529e-01 7.90232778e-01 9.99156952e-01]
  [3.98981631e-01 7.82709479e-01 7.35594213e-01 9.70018148e-01]
  [6.31434858e-01 6.56596124e-02 9.94844258e-01 9.91125226e-01]
  [4.03546631e-01 3.72542024e-01 8.93805206e-01 7.11475253e-01]
  [4.92211223e-01 2.52640486e-01 7.18696833e-01 4.07687068e-01]
  [6.85503840e-01 4.34535980e-01 8.68745089e-01 4.70350862e-01]
  [3.38426471e-01 5.05429626e-01 3.66816342e-01 5.16116381e-01]
  [5.56789041e-01 0.00000000e+00 8.72881413e-01 6.77806586e-02]
  [7.46141136e-01 4.60771352e-01 8.83682549e-01 5.43849111e-01]
  [3.97113353e-01 4.10500914e-01 4.45086926e-01 4.24319774e-01]
  [3.31506431e-01 4.76335794e-01 3.56430948e-01 4.88202840e-01]
  [5.03257632e-01 2.41804078e-01 6.15983486e-01 3.58820289e-01]
  [4.19946134e-01 1.86517924e-01 9.83408988e-01 9.87705827e-01]
  [5.36903977e-01 2.56856382e-01 6.54782653e-01 3.43218327e-01]
  [4.12004054e-01 5.86636543e-01 6.93381369e-01 1.00000000e+00]
  [5.20348430e-01 2.73481667e-01 6.10880256e-01 3.93006027e-01]
  [5.74276567e-01 2.87748992e-01 7.21984267e-01 3.91383827e-01]
  [2.89862007e-01 2.84560740e-01 7.39836216e-01 6.37443185e-01]
  [4.96246338e-01 2.63413846e-01 7.70015240e-01 4.57567930e-01]
  [5.19865811e-01 1.98258713e-01 6.03035390e-01 2.73524016e-01]
  [4.83228505e-01 2.16619462e-01 6.39268816e-01 4.49601918e-01]
  [3.84083927e-01 4.67547536e-01 4.31649745e-01 4.81992364e-01]
  [3.48392308e-01 3.74274254e-01 5.03999054e-01 6.12540483e-01]
  [5.74921966e-01 3.47678363e-03 9.15780425e-01 2.38593623e-01]
  [3.45435590e-01 5.46659231e-01 3.74062091e-01 5.60205579e-01]
  [6.14662647e-01 1.25400722e-02 9.80371952e-01 4.75324601e-01]
  [6.86917663e-01 4.48385119e-01 8.99734139e-01 6.21741176e-01]
  [5.43244362e-01 2.79347658e-01 6.78029895e-01 3.91894460e-01]
  [3.64326298e-01 0.00000000e+00 6.73295557e-01 7.37453580e-01]
  [6.07542276e-01 5.06267175e-02 7.68457413e-01 2.98367083e-01]
  [2.58686990e-01 4.09406722e-02 7.81508684e-01 6.48879886e-01]
  [4.82539684e-01 2.30819955e-01 5.28611958e-01 3.63296390e-01]
  [4.15112287e-01 7.37580359e-01 7.52923012e-01 9.21294510e-01]
  [3.98340166e-01 4.32618022e-01 7.49099314e-01 6.08014822e-01]
  [5.01145720e-01 3.60702038e-01 6.23760223e-01 4.31556940e-01]
  [4.80772495e-01 2.87768871e-01 5.28597713e-01 3.74635667e-01]
  [7.58994102e-01 5.31462878e-02 8.83683920e-01 3.54582548e-01]
  [7.33969390e-01 4.40615416e-01 8.75839055e-01 4.80525196e-01]
  [3.32216144e-01 4.08021301e-01 5.61710179e-01 5.81104398e-01]
  [5.27267098e-01 2.43273854e-01 8.85313392e-01 4.95633841e-01]
  [6.36232436e-01 1.87747717e-01 8.11940730e-01 2.77306348e-01]
  [5.35063982e-01 3.03871930e-04 9.76815224e-01 1.71860784e-01]
  [5.52567542e-01 3.21750902e-03 6.88723385e-01 6.52680770e-02]
  [5.18970549e-01 1.09627441e-01 8.73319566e-01 4.07769084e-01]
  [4.70146358e-01 3.06456447e-01 6.97903335e-01 4.66317654e-01]
  [4.12567884e-01 2.80987918e-01 8.80960703e-01 6.27162635e-01]
  [5.88587642e-01 2.44935393e-01 8.71819854e-01 3.90277863e-01]
  [5.04896283e-01 5.80270052e-01 7.68337965e-01 1.00000000e+00]
  [4.37421232e-01 3.08278799e-01 7.68657207e-01 7.64984012e-01]
  [5.28930783e-01 3.55813205e-01 8.43232036e-01 5.18654346e-01]
  [7.06788957e-01 4.37110692e-01 9.14967120e-01 5.72017550e-01]
  [2.18201816e-01 4.88946289e-01 8.62469971e-01 9.63312864e-01]
  [6.22738302e-01 1.38248444e-01 8.57222259e-01 2.82573849e-01]
  [6.54138923e-01 3.04315478e-01 7.45888233e-01 3.82857710e-01]
  [4.29467261e-01 7.80698359e-01 6.48266017e-01 1.00000000e+00]
  [4.46878254e-01 2.09028199e-01 6.40358984e-01 3.75346243e-01]
  [3.16258848e-01 8.32504749e-01 8.23982418e-01 9.90386963e-01]
  [6.05024457e-01 2.52521902e-01 8.12417507e-01 4.83819872e-01]
  [3.51336688e-01 4.98775810e-01 1.00000000e+00 9.06152248e-01]
  [6.95564687e-01 7.50948310e-01 8.88447940e-01 1.00000000e+00]
  [3.88258576e-01 5.33289671e-01 8.75454664e-01 9.59217429e-01]
  [3.02734107e-01 5.05446017e-01 3.32951814e-01 5.21313846e-01]
  [5.78578115e-01 2.70111322e-01 7.26399779e-01 3.33653688e-01]
  [8.31338286e-01 2.76189297e-02 9.03123498e-01 3.93661559e-01]
  [5.81698239e-01 1.67757824e-01 8.35726202e-01 2.95289814e-01]
  [5.36950946e-01 3.19106877e-02 9.08833861e-01 7.96007037e-01]
  [3.30216944e-01 1.85503408e-01 6.03329062e-01 6.20327592e-01]
  [6.18818402e-01 0.00000000e+00 7.89601564e-01 1.48627713e-01]
  [6.30266726e-01 4.33205903e-01 8.52360308e-01 5.72274148e-01]
  [3.35949063e-01 4.14158136e-01 3.57951701e-01 4.26587313e-01]
  [7.20640838e-01 8.04916024e-03 9.58042920e-01 5.61199069e-01]
  [5.84677398e-01 7.03474283e-02 7.36290991e-01 3.44596863e-01]
  [4.19883460e-01 4.06788111e-01 4.68255132e-01 4.25109506e-01]
  [6.39927804e-01 2.00164318e-03 9.97901261e-01 8.11412573e-01]
  [5.33949196e-01 2.72342533e-01 8.78523052e-01 4.32817489e-01]
  [6.58142984e-01 4.44802016e-01 7.81714380e-01 5.50549507e-01]
  [3.07564199e-01 3.41288984e-01 6.50976717e-01 6.01556361e-01]
  [4.40935194e-01 1.88440084e-02 9.83413517e-01 9.33977723e-01]
  [4.89288419e-01 0.00000000e+00 1.00000000e+00 1.12769425e-01]
  [7.88989484e-01 2.11772084e-01 1.00000000e+00 3.67680490e-01]
  [5.96338212e-01 3.88508767e-01 6.47209346e-01 4.74365383e-01]
  [3.38769794e-01 3.37847918e-01 8.24353933e-01 5.76343238e-01]
  [4.90569234e-01 1.95525885e-01 7.80579686e-01 3.27015400e-01]
  [5.47312975e-01 3.57423306e-01 6.47958755e-01 4.23456550e-01]
  [6.18103385e-01 6.21652603e-02 8.44633698e-01 2.22137511e-01]
  [5.27049541e-01 1.17950372e-01 6.17353082e-01 1.53708220e-01]
  [7.16908813e-01 4.01781499e-03 9.84125972e-01 3.00909281e-01]
  [5.44539571e-01 2.41813436e-03 8.41970801e-01 1.26415297e-01]
  [4.44469690e-01 8.28753531e-01 7.46623278e-01 9.87598717e-01]
  [3.15123707e-01 4.62584257e-01 3.35273296e-01 4.77230787e-01]
  [3.47529173e-01 3.27103466e-01 7.68559933e-01 9.83226418e-01]
  [4.02757078e-01 3.62969428e-01 4.32684571e-01 3.75950783e-01]
  [7.05794036e-01 1.78262621e-01 9.70153272e-01 2.98915476e-01]
  [5.08592129e-01 7.52241552e-01 7.34334707e-01 9.93464291e-01]
  [1.42102808e-01 0.00000000e+00 4.66604441e-01 7.86211610e-01]
  [5.85047305e-01 1.92644715e-01 6.23101652e-01 2.79952288e-01]
  [6.69153452e-01 7.77340114e-01 7.54663706e-01 8.55158150e-01]
  [3.93155396e-01 9.21279490e-01 7.05600083e-01 9.97392118e-01]]]
Counter({'person': 1, 'chair': 1, 'backpack': 1}) 2018-09-20 17:07:59.300454
Improve this question
3
  • Do you get a solution for it? I have also a similar requirement. If you got a solution please share it Commented Feb 7, 2019 at 9:00
  • @KK Nope I didn't found it but I have found a work around for my problem. Commented Feb 7, 2019 at 9:03
  • can you please share it Commented Feb 8, 2019 at 7:59

1 Answer 1

Reset to default
2

You can try this

image = Image.open(image_path)
# the array based representation of the image will be used later in order to prepare the
# result image with boxes and labels on it.
image_np = load_image_into_numpy_array(image)  
# Expand dimensions since the model expects images to have shape: [1, None, None, 3]
image_np_expanded = np.expand_dims(image_np, axis=0)
# Actual detection.
output_dict = run_inference_for_single_image(image_np, detection_graph)

width = 1024
height = 600
threshold = 0.5

for i,j in zip(output_dict['detection_boxes'],output_dict['detection_scores']):
    if(j>threshold):
        print(i[1]*width,i[0]*height,i[3]*width,i[2]*height)

This will output the boxes in the order of xmin,ymin,xmax,ymax. I took 0.5 as the min_score_threshold. Hope this helped you. Please let me know in case you face any issues

Improve this answer
Sign up to request clarification or add additional context in comments.

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.