In detectron2 the output get the bellow output upon predection.
panoptic_seg, segments_info = predictor(im)["panoptic_seg"]
it can be drawn via draw_panoptic_seg_predictionsfunction, but we want to save the mask deatils & redraw using opencv how can we do that ?
for instance segmentation we can do like this :
import cv2
import numpy as np
# "outputs" is the inference output in the format described here - https://detectron2.readthedocs.io/tutorials/models.html#model-output-format
# Extract the contour of each predicted mask and save it in a list
contours = []
for pred_mask in outputs['instances'].pred_masks:
# pred_mask is of type torch.Tensor, and the values are boolean (True, False)
# Convert it to a 8-bit numpy array, which can then be used to find contours
mask = pred_mask.to("cpu").numpy().astype('uint8')
contour, _ = cv2.findContours(mask, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)
contours.append(contour[0]) # contour is a tuple (OpenCV 4.5.2), so take the first element which is the array of contour points
and then draw the mask with this code
# "image" is the original BGR image
image_with_overlaid_predictions = im.copy()
for contour in contours:
cv2.drawContours(image_with_overlaid_predictions, [contour], -1, (0,255,0), 20)
plt.figure()
plt.imshow(image_with_overlaid_predictions)
I want to achive similar with pantopic mask
Related
I am trying to get diameters on different points of a cylinder over a certain length using computer vision to replace the use of optical micrometer.
Image of a cylinder:
How can I calculate the diameter of this object (cylinder) on multiple points (blue lines) along its length as shown in the image using OpenCV python?
An OpenCV solution. The main idea is to:
Detect edges
Find the contours of the edges
Fill in the contour areas
Go through each column in the image and count the nonzero pixels
1., 2. and 3. could possibly be simplified by a single thresholding step depending on your use case
import numpy as np
import cv2
src = cv2.imread('/path/to/src.jpg')
mask = np.zeros(src.shape, dtype=np.uint8)
w, h, c = src.shape
# edge detection
threshold = 100
gray = cv2.Canny(src, threshold, threshold * 2)
cv2.imshow('', gray)
cv2.waitKey(0)
# find contours
cnts = cv2.findContours(gray, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
threshold_area = 0.5
# fill area withing contours with white color
for c in cnts:
area = cv2.contourArea(c)
if area > threshold_area:
cv2.drawContours(mask, [c], -1, (255, 255, 255), -1)
cv2.imshow('', mask)
cv2.waitKey(0)
# get non zero values (height) of each column
column_pixels = [cv2.countNonZero(mask[:, i]) for i in range(0, w)]
print(column_pixels)
Src image:
Canny result:
After filling in contours with white color:
countNonZero is applied on this last image for each column
I have an IR image with a resolution of (240 x 320), datatype: float32 as you can see here:
The raw npy image is here.
My objective is to detect the table (brown contour) in order to crop this region as a ROI.
What I have tried so far is to do:
Histogram equalization to increase contrast,
Gaussian Blurring to reduce the noise, and
contour detection to find the rectangular table in the middle of the image.
Note that the table in my case is installed on wheels, and hence it might slightly move so I want to detect it dynamically, and not use its fixed position inside the image.
The code I have used is:
#!/usr/bin/python3
# -*- coding: utf-8 -*-
import numpy as np
import matplotlib.pyplot as plt
import cv2
import random as rng
path = ""
# Read the numpy array
ir_raw = np.load(path+"ir.npy") # (240, 320) float32
ir = np.uint8((ir_raw/ir_raw.max()) * 255)
# Histogram equalization (Contrast Adjustment)
heq_ir = cv2.equalizeHist(ir)
# Gaussian smoothing (Noise Removal)
g_blur_ir = cv2.GaussianBlur(heq_ir, (5,5), 0)
# Otsu Thresholding
_, thresh_ir = cv2.threshold(g_blur_ir, 120, 255, cv2.THRESH_BINARY +
cv2.THRESH_OTSU)
# Find contours
contours, hierarchy = cv2.findContours(thresh_ir, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
# Draw contours
rng.seed(12345)
drawing = np.zeros((thresh_ir.shape[0], thresh_ir.shape[1], 3), dtype=np.uint8)
for i in range(len(contours)):
color = (rng.randint(0,256), rng.randint(0,256), rng.randint(0,256))
cv2.drawContours(drawing, contours, i, color, 2, cv2.LINE_8, hierarchy, 0)
plt.subplot(121)
plt.imshow(heq_ir)
plt.title("IR")
plt.subplot(122)
plt.imshow(drawing)
plt.title("IR contours")
plt.show()
Can you please tell me how can I detect the rectangular table in order to crop it as a ROI? thanks in advance.
Assuming the following, the table always has same area (well duh) but in the case below its always the largest object detected (the code should be easily to amend to filter out objects of different areas though).
I've filtered out all contours that have a different area to the table contour (process of elimination until I found the table then set the area threshold accordingly) & then fitted a rectangle to that contour. It is possible to fit skewed rectangles in OpenCV as well but I havent in this case. (EDIT: code for skewed rectangles added)
Your ROI boundary is in boundRect
for i in range(len(contours)):
if cv2.contourArea(contours[i]) > 10000:
color = (rng.randint(0,256), rng.randint(0,256), rng.randint(0,256))
cv2.drawContours(drawing, contours, i, color, 2, cv2.LINE_8, hierarchy, 0)
# For unrotated bounding rectangle
boundRect = cv2.boundingRect(contours[i])
cv2.rectangle(drawing, (int(boundRect[0]), int(boundRect[1])), (int(boundRect[0]+boundRect[2]), int(boundRect[1]+boundRect[3])), (255, 255, 255), 2)
# For minimal bounding rectangle that will rotate with table rotation
rect = cv2.minAreaRect(contours[i])
box = cv2.boxPoints(rect)
box = np.int0(box)
cv2.drawContours(drawing, [box], 0, (255, 0, 255), 2)
print(cv2.contourArea(contours[i]))
Ouput,
I have been trying to remove every other component from the image except the beans. I have tried using edging and contour I couldn't get it right.
enter image description here
If the "beans" are the dark region, then here is one way in Python OpenCV.
- Read the input
- Threshold on the blue color
- Apply morphology close to clean it up a little
- Invert
- Find the largest contour
- Draw a white filled contour on a black background as a mask
- Use the mask to make everything in the input black except the "beans"
- Save the results
Input:
import cv2
import numpy as np
# load image
img = cv2.imread("beans.jpg")
lower =(80,70,30)
upper = (220,220,180)
# create the mask and use it to change the colors
thresh = cv2.inRange(img, lower, upper)
# apply morphology
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5,5))
morph = cv2.morphologyEx(thresh, cv2.MORPH_CLOSE, kernel)
# invert
morph = 255 - morph
# find largest contour
contours = cv2.findContours(morph, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
contours = contours[0] if len(contours) == 2 else contours[1]
big_contour = max(contours, key=cv2.contourArea)
# draw white filled contour on black background as mask
mask = np.zeros_like(thresh)
cv2.drawContours(mask, [big_contour], 0, 255, -1)
# apply mask to img
result = img.copy()
result[mask==0] = (0,0,0)
# write result to disk
cv2.imwrite("beans_thresh2.png", thresh)
cv2.imwrite("beans_morph2.png", morph)
cv2.imwrite("beans_mask2.png", mask)
cv2.imwrite("beans_result2.png", result)
# display it
cv2.imshow("thresh", thresh)
cv2.imshow("morph", morph)
cv2.imshow("mask", mask)
cv2.imshow("result", result)
cv2.waitKey(0)
cv2.destroyAllWindows()
Threshold image:
Morphology cleaned image:
Mask Image:
Result Image:
I was successful with tensorflow object detection API like an image below.
but now I want to read character inside green box, how to do that?
First you need to crop out the bounding boxes of plate(s) and then you can use tesseract. To get the text. A pseudo code for your problem may look like:
import cv2
import pytesseract
original_img = cv2.imread("/path/to/your/img.png")
for plate in detected_plates:
if plate.confidence > 0.98:
b_box = plate.bounding_rect # bounding box in [x, y, w, h]
img_cropped = original_img[b_box[1]: b_box[1] + b_box[3], b_box[0]: b_box[0] + b_box[2]]
print(pytesseract.image_to_string(img_cropped))
# Load image using OpenCV and
# expand image 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
image = cv2.imread(PATH_TO_IMAGE)
image_rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
image_expanded = np.expand_dims(image_rgb, 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: image_expanded})
# Draw the results of the detection (aka 'visulaize the results')
vis_util.visualize_boxes_and_labels_on_image_array(
image,
np.squeeze(boxes),
np.squeeze(classes).astype(np.int32),
np.squeeze(scores),
category_index,
use_normalized_coordinates=True,
line_thickness=1,
min_score_thresh=0.60)
# All the results have been drawn on image. Now display the image.
cv2.imshow('Object detector', image)
I would like to create a program that is able to extract lines from a graph.
For example, if a graph like this is inputted, I would just want the red line to be outputted.
Below I have tried to do this using a hough line transformation, however, I do not get very promising results.
import cv2
import numpy as np
graph_img = cv2.imread("/Users/2020shatgiskessell/Desktop/Graph1.png")
gray = cv2.cvtColor(graph_img, cv2.COLOR_BGR2GRAY)
kernel_size = 5
#grayscale image
blur_gray = cv2.GaussianBlur(gray,(kernel_size, kernel_size),0)
#Canny edge detecion
edges = cv2.Canny(blur_gray, 50, 150)
#Hough Lines Transformation
#distance resoltion of hough grid (pixels)
rho = 1
#angular resolution of hough grid (radians)
theta = np.pi/180
#minimum number of votes
threshold = 15
#play around with these
min_line_length = 25
max_line_gap = 20
#make new image
line_image = np.copy(graph_img)
#returns array of lines
lines = cv2.HoughLinesP(edges, rho, theta, threshold, np.array([]),
min_line_length, max_line_gap)
for line in lines:
for x1,y1,x2,y2 in line:
cv2.line(line_image,(x1,y1),(x2,y2),(255,0,0),2)
lines_edges = cv2.addWeighted(graph_img, 0.8, line_image, 1, 0)
cv2.imshow("denoised image",edges)
if cv2.waitKey(0) & 0xff == 27:
cv2.destroyAllWindows()
This produces the output image below, which does not accurately recognize the graph line. How might I go about doing this?
Note: For now, I am not concerned about the graph titles or any other text.
I would also like the code to work for other graph images aswell, such as:
etc.
If the graph does not have many noises around it (like your example) I would suggest to threshold your image with Otsu threshold instead of looking for edges . Then you simply search the contours, select the biggest one (graph) and draw it on a blank mask. After that you can perform a bitwise operation on image with the mask and you will get a black image with the graph. If you like the white background better, then simply change all black pixels to white. Steps are written in the example. Hope it helps a bit. Cheers!
Example:
import numpy as np
import cv2
# Read the image and create a blank mask
img = cv2.imread('graph.png')
h,w = img.shape[:2]
mask = np.zeros((h,w), np.uint8)
# Transform to gray colorspace and threshold the image
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
# Search for contours and select the biggest one and draw it on mask
_, contours, hierarchy = cv2.findContours(thresh,cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)
cnt = max(contours, key=cv2.contourArea)
cv2.drawContours(mask, [cnt], 0, 255, -1)
# Perform a bitwise operation
res = cv2.bitwise_and(img, img, mask=mask)
# Convert black pixels back to white
black = np.where(res==0)
res[black[0], black[1], :] = [255, 255, 255]
# Display the image
cv2.imshow('img', res)
cv2.waitKey(0)
cv2.destroyAllWindows()
Result:
EDIT:
For noisier pictures you could try this code. Note that different graphs have different noises and may not work on every graph image since the denoisiation process would be specific in every case. For different noises you can use different ways to denoise it, for example histogram equalization, eroding, blurring etc. This code works well for all 3 graphs. Steps are written in comments. Hope it helps. Cheers!
import numpy as np
import cv2
# Read the image and create a blank mask
img = cv2.imread('graph.png')
h,w = img.shape[:2]
mask = np.zeros((h,w), np.uint8)
# Transform to gray colorspace and threshold the image
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
# Perform opening on the thresholded image (erosion followed by dilation)
kernel = np.ones((2,2),np.uint8)
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel)
# Search for contours and select the biggest one and draw it on mask
_, contours, hierarchy = cv2.findContours(opening,cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)
cnt = max(contours, key=cv2.contourArea)
cv2.drawContours(mask, [cnt], 0, 255, -1)
# Perform a bitwise operation
res = cv2.bitwise_and(img, img, mask=mask)
# Threshold the image again
gray = cv2.cvtColor(res,cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)
# Find all non white pixels
non_zero = cv2.findNonZero(thresh)
# Transform all other pixels in non_white to white
for i in range(0, len(non_zero)):
first_x = non_zero[i][0][0]
first_y = non_zero[i][0][1]
first = res[first_y, first_x]
res[first_y, first_x] = 255
# Display the image
cv2.imshow('img', res)
cv2.waitKey(0)
cv2.destroyAllWindows()
Result: