def segmenting_images(self):
self.log.info('Loading model...')
segment_image = custom_segmentation()
segment_image.inferConfig(num_classes=1, class_names=["PigFace"], detection_threshold=0.95)
# segment_image.load_model("model/model.h5")
segment_image.load_model("../model/mask_rcnn_model.036-0.139239.h5")
# segment_image.segmentImage("1.png", "../app/upload/1.png" , show_bboxes=True, output_image_name="../output/1.png", verbose=True)
dir_path = project_config.image_upload_dir_path
output_path = project_config.output_dir_path
while True:
self.log.info('Ready to upload directory', datetime.now())
i = 0
files = [x for x in os.listdir(dir_path) if (x.endswith('.jpg') or x.endswith('.JPG') or x.endswith('.png') or x.endswith('.PNG'))]
for imageFullFileName in files:
image_file_name = os.path.basename(imageFullFileName)
self.log.info('Image filename: ', image_file_name)
start_time = datetime.now()
try:
segment_image.segmentImage(image_file_name, dir_path + r"/" + image_file_name, show_bboxes=True, output_image_name=output_path + r"/" + image_file_name)
except:
self.log.info('Image filename: ', image_file_name)
os.remove(os.path.join(dir_path, imageFullFileName))
end_time = datetime.now()
diff = (end_time-start_time).microseconds / 1000
self.log.info('Elapsed time for Segmentation: ', '%.2gs' % diff)
time.sleep(0.2)
def annotateSingle(input_path, output_path):
"""
Annotates a single image
Args:
input_path: The raw image file
output_path: The file where the annotated image should be stored
"""
segment_image = custom_segmentation()
segment_image.inferConfig(num_classes= 1, class_names= ["BG", ""])
img = cv2.imread(input_path)
h, w, colors = img.shape
grayscale_matrix = np.zeros((h, w))
for y in range(h):
for x in range(w):
rgb_values = img[y,x]
grayscale_matrix[y,x] = np.min(rgb_values)
if np.mean(grayscale_matrix) < 255/2:
model = DARK_MODEL
else:
model = LIGHT_MODEL
segment_image.load_model(model)
segment_image.segmentImage(input_path, show_bboxes=False, output_image_name=output_path,
extract_segmented_objects= False, save_extracted_objects=False)
def __init__(self,
resolution=(720, 800),
model_path=os.path.join(p().SEG_MODELS, 'D.h5'),
intrinsics='1280_720_color'):
self.master = custom_segmentation()
self.master.inferConfig(num_classes=1, class_names=["BG", "mh5"])
self.master.load_model(model_path)
self.crop_resolution = resolution
self.intrinsics = proj.makeIntrinsics(intrinsics)
def annotateFolder(input_path, output_path):
"""
Annotates an entire folder of worm images and stores it in a new folder
Uses LIGHT_MODEL and DARK_MODEL based on the average shade of each image.
Args:
input_path: The folder location storing the images relative to the python file
output_path: The folder location to store the annotated images
"""
all_image = os.listdir(input_path)
os.mkdir(output_path)
segment_light = custom_segmentation()
segment_light.inferConfig(num_classes= 1, class_names= ["BG", ""])
segment_light.load_model(LIGHT_MODEL)
segment_dark = custom_segmentation()
segment_dark.inferConfig(num_classes= 1, class_names= ["BG", ""])
segment_dark.load_model(DARK_MODEL)
for item in all_image:
img = cv2.imread(input_path + "/"+item)
h, w, colors = img.shape
grayscale_matrix = np.zeros((h, w))
for y in range(h):
for x in range(w):
rgb_values = img[y,x]
grayscale_matrix[y,x] = np.min(rgb_values)
if np.mean(grayscale_matrix) > 255/2:
print("Light")
segment_light.segmentImage(input_path+"/"+item, show_bboxes=False, output_image_name=output_path+"/Annotated_"+item,
extract_segmented_objects= False, save_extracted_objects=False)
else:
print("Dark")
segment_dark.segmentImage(input_path+"/"+item, show_bboxes=False, output_image_name=output_path+"/Annotated_"+item,
extract_segmented_objects= False, save_extracted_objects=False)
import pixellib
from pixellib.instance import custom_segmentation
test_video = custom_segmentation()
test_video.inferConfig(num_classes=7,
class_names=[
"BG", "base_link", "link_s", "link_l", "link_u",
"link_r", "link_b"
])
test_video.load_model("models/segmentation/multi/A.h5")
test_video.process_video("data/set10/og_vid.avi",
show_bboxes=True,
output_video_name="output/multiseg_test.avi",
frames_per_second=15)
import pixellib
from pixellib.instance import custom_segmentation
import cv2
capture = cv2.VideoCapture(0)
segment_camera = custom_segmentation()
segment_camera.inferConfig(num_classes=2,
class_names=["BG", "butterfly", "squirrel"])
segment_camera.load_model("Nature_model_resnet101.h5")
segment_camera.process_camera(capture,
frames_per_second=10,
output_video_name="output_video.mp4",
show_frames=True,
frame_name="frame")
# in this lesson we would learn how to use our model , and detect sevral objects in an image
# Moreover , we will extract the object to seperated images
import pixellib
import cv2
import numpy as np
from pixellib.instance import custom_segmentation
segment_image = custom_segmentation()
segment_image.inferConfig(network_backbone="resnet101",
num_classes=3,
class_names=["BG", "Banana", "Apple", "Tomato"])
segment_image.load_model("c:/models/eval/mask_rcnn_model.032-0.200773.h5")
segmask, output = segment_image.segmentImage(
"C:/GitHub/Object-Detection/Pixellib/moreThanOneApple.jpg",
show_bboxes=True,
extract_segmented_objects=True,
save_extracted_objects=True,
output_image_name=
"C:/GitHub/Object-Detection/Pixellib/moreThanOneAppleOut.jpg")
# lets see the object
#print(segmask)
# We need this section : extracted_objects
res = segmask["extracted_objects"]
#firstImage = res[0]
#firstImageArr = np.uint8(firstImage)