A. Nasirzadeh Dashti - University of Mohaghegh Ardebili,

Objects detection in images is a computer vision task with numerous practical applications. There are many ways to do this, but we need methods that are easy to use in different applications. The purpose of this paper is not to present a complex objects detector that has been trained in a massive dataset. this paper wants to demonstrate the U-Net capability that, with modifications to it, achieves to maximum detection accuracy with fewer parameters and shorter training times in a small data set. As a result, because of its ease of use and high precision can be Consideration in industrial and robotic applications. Classical U-Net architectures is very popular for segmentation of medical images and satellite images, but we use it as an object detector and for segmentation too in image of cluttered environment. For this paper, we created a dataset containing ۵۰۰ images from ۱۰ objects that are cluttered in each image. the model was evaluated on dataset and The results was impressive.

Object Detection, Image Segmentation, U-Net, Deep Convolutional Network, Computer Vision