Authors :
P. S. Ezekiel; O. E. Taylor; D. J. S. Sako
Volume/Issue :
Volume 6 - 2021, Issue 7 - July
Google Scholar :
http://bitly.ws/9nMw
Scribd :
https://bit.ly/2XnSO89
Abstract :
Road potholes are extensively enormous
primary disappointments out and about surface. They
are brought about by withdrawal and extension of the
road surface as water saturates into the ground. To
guarantee traffic wellbeing, it is critical and important to
regularly investigate and fix road potholes. This paper
presents a smart system for potholes detection using
computer vision with transfer learning. The system
starts by acquiring a pothole images as data, preprocessing the data by creating image annotation and
image augmentation on the pothole images. We also
created sub directories by creating called images and
annotation where we placed our training images in the
image folder and the annotated files (which was
generated in Xml format) into the annotation folder that
we created. We then train our model using transfer
learning. By transfer learning, we downloaded a pretrained yolov3 weight file trained on a coco dataset for
object detection. We then set the batch_size to be equal
to 4, with a tensorflow Gpu of version ==1.13.1, number
of experiments = 200, and
train_from_pre_trained_model = pre-trained-yolov3
(the weight file we downloaded). After training, we
evaluated and saved the trained model. We had a
training accuracy of about 97%. We carried out a real
time pothole detection on a live streaming video using
opencv library, where we detected multiple potholes
images.
Keywords :
Potholes, Computer Vision, Transfer Learning, Yolov3.
Road potholes are extensively enormous
primary disappointments out and about surface. They
are brought about by withdrawal and extension of the
road surface as water saturates into the ground. To
guarantee traffic wellbeing, it is critical and important to
regularly investigate and fix road potholes. This paper
presents a smart system for potholes detection using
computer vision with transfer learning. The system
starts by acquiring a pothole images as data, preprocessing the data by creating image annotation and
image augmentation on the pothole images. We also
created sub directories by creating called images and
annotation where we placed our training images in the
image folder and the annotated files (which was
generated in Xml format) into the annotation folder that
we created. We then train our model using transfer
learning. By transfer learning, we downloaded a pretrained yolov3 weight file trained on a coco dataset for
object detection. We then set the batch_size to be equal
to 4, with a tensorflow Gpu of version ==1.13.1, number
of experiments = 200, and
train_from_pre_trained_model = pre-trained-yolov3
(the weight file we downloaded). After training, we
evaluated and saved the trained model. We had a
training accuracy of about 97%. We carried out a real
time pothole detection on a live streaming video using
opencv library, where we detected multiple potholes
images.
Keywords :
Potholes, Computer Vision, Transfer Learning, Yolov3.
