• Fatima M Inamdar
• Bhavesh Ingale
• Nidhiy Deshpande
• Shrirang Deshpande
• Rushikesh Gobare

Dog Breed Identification is a model used for detecting the breed of the image of the dog provided. It is a classification category project where various images of dogs are classified based on their breeds. There are over 120 breeds of dogs that can be classified in the project. The images classification is one of the most common applications of deep learning. Images of dogs are mostly used as examples for image classification models, as they are relatively easy for the human eyes to recognize. The paper can give a model which can classify dog breeds. So, it can be used in various scenarios. A pet care application can be a potential use case of the model. The algorithm can be used to be deployed in an application to classify dog breeds. This model can also be used by pet enthusiasts to know a breed of a certain dog. Various NGOs are specially dedicated to care for the stray and the abandoned dogs. These NGO’s can use this model on the daily base since availability of veterinary doctors is limited. An identifier like this could be used in bio-diversity studies, helping scientists save time and resources when conducting studies about the health and abundance of certain species populations. These studies are crucial for assessing the status of ecosystems, and accuracy during these studies is particularly important because of their influence on policy changes. Breed prediction may also help veterinarians treat breed specific ailments for stray, unidentified dogs that need medical care. Ultimately, we found dogs to be the most interesting class to experiment with due to their immense diversity, loving nature, and abundance in photographs, but we also hope to expand our understanding of the fine-grained classification problem and provide a useful tool for scientists across disciplines. This work was implemented using tensor flow where a number of dog images with their breeds were provided as data set. The source of the data set was Kaggle. We will use data set from Kaggle. The data set contains 10000+ images with 120 different breeds of dogs. The data set will contain null or non-acceptable values (values which are not present in required format) which have to be removed or treated accordingly. The treatment can be done by checking if the data of a particular category are mean based or median based. And then the subsequent values can be utilized. Then using the modified data set we can train our model for the data set. For this step we will be using TensorFlow. TensorFlow is an end-to-end machine learning solution. Now, we can create a tensor of a specific shape to represent various features of data. The model now needs a definition of various Keras sequential layers. Now the model will be built on these layers. This research work will use test data set for the evaluation of model. Tensor Board metrics will be used to evaluate our created model. And precautions should be taken that an overfit or underfit is avoided. If the metric shows too high accuracy the model has adapted to the training data set and it is an overfit. If the metric shows too less accuracy the model needs more training and it is an underfit. Then using this model, will test on data set.

Dog Breed Classification, Image Classification, Artificial Intelligence, Keras.

"DOG BREED IDENTIFICATION MODEL USING ARTIFICIAL INTELLIGENCE", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 5, page no.n51-n55, May 2023, Available :http://www.jetir.org/papers/JETIR2305D10.pdf

"DOG BREED IDENTIFICATION MODEL USING ARTIFICIAL INTELLIGENCE", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 5, page no. ppn51-n55, May 2023, Available at : http://www.jetir.org/papers/JETIR2305D10.pdf