In recent years more and more focus has been placed on the environmental aspects of ships, because of the great attention to exhaust gas emissions from ships including CO2, a leading contributor to Green House Gasses (GHG) emissions, due to their negative effect on global warming. Maritime transport has clear environmental advantages: it expends relatively little energy and its infrastructure requirements are small compared to land-based transport modes. Due to low energy need, shipping is a highly carbon-efficient transport mode, i.e. carbon dioxide emissions are low compared to the weight of cargo transported. Shipping can be up to four times more efficient than road transport. Because of relatively small contribution to greenhouse gas emissions shipping is also good in the terms of mitigation of climate change. However, air pollution from ships has been unregulated until recently. Ships currently produce about half as much sulphur oxides (SOx ) as land-based sources and about a third as much nitrogen oxides (NOx ). Ships emit several hazardous air pollutants such as sulphur dioxide, nitrogen oxides and fine particles. Once emitted, airborne emissions can travel considerable distances so the shipping emissions affect land air quality. Also the emissions from ships during port stays can be substantial contributor to the local air quality. Thus,this report describes the different exhaust gas emission products, their untoward effects and the various ways in which it can be mitigated

Deep learning these days is playing a very major role in natural signal and information processing,like speech recognition and image classification. Deep learning technologyis immersed on the basis of the human brain. In deep learning, artificial neurons network automaticallytrained itself by large datasets that discovers connected patterns without the help of a human.Deep learning detects a pattern in unstructured data like image, sound, video, and text.For Image classification, CNN in deep learning is very popular. In many patterns, CNN performs bettercompared to human in a large dataset like image. In our research, we have used python with Kerasfor binary image classification. In this, we are using an animal’s dataset, namely cat and dog, for image classification. Four different parameters with four different combinations have been appliedin CNN for comparison. It is shown that for Binary image classification combination of sigmoid classifier and Relu activation function gives higher classification accuracy than any other combination of classifier and activation function