Underwater image processing has always been a promising and thrilling task due to the natural condition and the lighting effect for taking the image requires good artificial lights. While taking underwater images lots of difficulties are faced by photographers such as the shadows, non-uniform lighting, color shading, etc. Recognizing the object underwater is very difficult in order to the environmental condition. Man-made object recognition was made with underwater optical sensors to capture underwater images that have gained more attention from the users. Deep learning methods have demonstrated impressive performance in object recognition tasks from natural images. Anyhow it is hard to collect all the labelled underwater optical images for training the model. It is possible to acquire labelled images. Based on the assumption that it is possible to acquire sufficient labeled in-air images, the proposed work leverages a combination of deep learning and transfer learning to develop a novel recognition system for the man-made object from underwater optical images. The extracted features from the proposed network have high representative power and demonstrate robustness in both in-air and underwater imaging modalities. Therefore, our proposed framework has the ability to recognize underwater man-made objects using only labeled in-air images. The results of experiments on simulated data demonstrate that the proposed method outperforms traditional deep learning methods in the task of underwater man-made object recognition.