Abstract: Some, if not all, veterinary clinics do not have a record of skin diseases of dogs when they diagnosed them; this is due to lack of manpower over the number of different kinds of animal patient that they cater per day. This also causes some delays in diagnosing other non-visually diseases that other patients might have. Having a system that can be used in pre-examination for visually available infections such as dog’s skin disease and automatically records this diagnosis, can give an advantage to veterinary clinics. This helps the clinics prepare for the kind of patients that they will tend in certain time of the year. Developing a system that can be used in identifying common dog skin diseases for the pre-examination purpose and creating a dashboard that generates a numerical result can be used as an advantage for the veterinary clinics. These can be achieved by using image processing techniques for the prediction model and Convolutional Neural Network (CNN). However, using a common CNN approach, where the main core uses the Central Processing Unit (CPU), tends to train the model longer. To overcome this problem, the use Graphics Processing Unit (GPU) is implemented to enhance the speed of training the model for the system. Having this kind of system really helps the veterinary clinic for their daily work, but this can still be improved by using other approaches to the trained model without ignoring the efficiency and accuracy of the algorithm that is being used.