Food Identification using Convolutional Neural Network with Pretrained models
In this project food identification is done through deep convolutional neural networks.
Platform : Matlab
Delivery : One Working Day
Support : Online Demo ( 2 Hours)
100 in stock
Food Identifcation using Deep learning Matlab
In this paper it is examined the effectiveness of Deep Convolutional Neural Network ( DCNN ) for recognizing the food . Food Identification is a kind of Visual recognition which is relatively harder. A best combination of Deep CNN related techniques which is Pre-training with the large set of dataset eg. ImageNet data and Fine tuning is done i.e transferring Learning from the pre-trained DCNN. A fine tuning is Done to the pre-trained model’s layer. After tuning the training is carried out with the food dataset and the food is identified. Finally the effectiveness of pre-trained model and fine-tuning of Deep CNN with small scale Food dataset is examined.
Food image Recognition is one of the Promising application of object recognition. A convolutional neural network can be used which is suited for image recognition. In this paper the effectiveness of Deep Convolutional Neural Network ( DCNN ) for recognizing the food which will solve the image classification problem. Machine Learning which involves using Complicated Models is called Deep Neural Network. It is a subset of Machine Learning in which Multilayered Neural Network learn from vast amount of data. Transfer learning is carried out be feeding a Pre-trained model to a new network Transfer Learning is done to the network to be created . Transfer Learning is that to use knowledge, that a model has Learned from a task where a lot of labeled training data is available. Instead of Starting the Learning Process from Scratch , it is started from the patterns that have been learned from solving the task which is related to it. A Pre-trained model which is trained with a large number of dataset in the network. This Transfer learning make it easier to create a network. The Imagenet Model is used in this Paper which is already trained with 14 million datasets.
A CNN is developed using already Pre-trained CNN Model , which is been trained on a dataset and a pretrained model’s is Loaded into the network . Food Identification is a kind of Visual recognition which is relatively harder. The Goal is to Recognize the food using a neural network. The neural network is trained with a large number of dataset. The network learns to predict the image and give which class it belongs to. The network is trained with different sets of labels which could predict different class of food. The layers are created depending upon the requirements. Here the image input layer is trained with 227 227. The last few layers of the Pre-Trained models network is removed and few more layer is added to it and the dataset is trained accordingly. After Transfer Learning the datasets are trained and food identification is carried out.
BLOCK DIAGRAM EXPLANATION
Initially the dataset is loaded for training . The pre-trained model which is stored in a .mat file has to be loaded . then a fine tuning is done to it. New layers are added to the Pre-trained network . the datasets are trained to the network which is created after fine tuning. The Convolutional neural network which is especially suited for image recognition is used here . the image is recognized after training.
Software : Matlab
Food identification is a very crucial step for estimation when it comes to lagre number of dataset. In this paper the effectiveness of Pre-trained model and fine tuning of Deep CNN with Food dataset is examined.
For more Image Processing projects ,Click here
For more Deep Learning Projects Click here