Targets classification of electro optical/infrared images offerings a needed task for observing and surveillance of
delicate areas such as military regions. Electro-optic (EO) image devices exhibit the characteristics of high determination and
low noise level at day, but it do not work in dark surroundings. Infrared (IR) image devices exhibition poor determination and
cannot separate target with comparable temperature. Therefore, an original context of IR image improvement depends on the
evidence from EO images, which recovers the determination of IR images and benefits to differentiate objects at night. Due to
the technology development there are different methods to overcome these challenges. One of the finest method is Machine
Learning Technique to train the model. Convolutional neural network and mask regional based CNN algorithms are used for
classification and finding the detection accuracy. In this project work data may be in the form of RGB images or binary images.
It take dataset up to 100 images
