0

But First: if anyone has a better way of getting screen capture into opencv, I'm all ears. this is just the way I've seen most people do it

I want to use a live screen recording to do object detection in opencv. I have no problems getting the video to display with

printscreen_pil = ImageGrab.grab() printscreen_numpy = np.array(printscreen_pil.getdata(),dtype='uint8')\ .reshape((printscreen_pil.size[1],printscreen_pil.size[0],3)) cv2.imshow('window',printscreen_numpy)

But I am running into type issues when I try to initialize object detection with `ret, frame = video.read()'

I get: 'AttributeError: 'numpy.ndarray' object has no attribute 'read'`

I have to assume that printscreen_numpy is in the wrong format, how do i convert it to a video that can be read by opencv?

This is where I got the code:

Screen Capture with OpenCV and Python-2.7

I have tried every combination of inserting the video into video.read() with no luck.

edit: meaning Ive tried:

printscreen_pil = ImageGrab.grab() ---> printscreen_pil.read() as well as: printscreen_pil = np.array(ImageGrab.grab()) and so forth

The relevant code block as it stands:

` while(True):

printscreen_pil =  ImageGrab.grab()
printscreen_numpy =   np.array(printscreen_pil.getdata(),dtype='uint8')\
.reshape((printscreen_pil.size[1],printscreen_pil.size[0],3))
cv2.imshow('window',printscreen_numpy) 

# Acquire frame and expand frame dimensions to have shape: [1, None, None, 3]
# i.e. a single-column array, where each item in the column has the pixel RGB value
ret, frame = printscreen_numpy.read()
frame_expanded = np.expand_dims(frame, axis=0)

# Perform the actual detection by running the model with the image as input
(boxes, scores, classes, num) = sess.run(
    [detection_boxes, detection_scores, detection_classes, num_detections],
    feed_dict={image_tensor: frame_expanded})

# Draw the results of the detection (aka 'visulaize the results')
vis_util.visualize_boxes_and_labels_on_image_array(
    frame,
    np.squeeze(boxes),
    np.squeeze(classes).astype(np.int32),
    np.squeeze(scores),
    category_index,
    use_normalized_coordinates=True,
    line_thickness=8,
    min_score_thresh=0.60)

# All the results have been drawn on the frame, so it's time to display it.
cv2.imshow('Object detector', frame)

# Press 'q' to quit
if cv2.waitKey(1) == ord('q'):
    break

`

Cory Future
  • 11
  • 3
  • you need to do some another object .read() cause numpy.ndarray dont have that method, afaik and as the error there is – user8426627 Jun 08 '19 at 02:21
  • but how do i convert from `numpy.ndarray` to an object that can be read with `.read()` ? – Cory Future Jun 08 '19 at 02:26
  • That doesn't make sense. Why do you even want to do `ret, frame = video.read()`? You already have the frame in your variable `printscreen_numpy`. – Mark Setchell Jun 08 '19 at 07:32
  • And why do you do that horrible thing with `getdata()`? You should be able to just do `printscreen_numpy = np.array(printscreen_pil)`. – Mark Setchell Jun 08 '19 at 07:33
  • Is the speed (frames/second) of capturing the screen important to you? If so, what rate are you achieving and what machine and OS are you using? – Mark Setchell Jun 08 '19 at 13:21

0 Answers0