The process of learning good features for machine learning applications can be very computationally expensive and may prove difficult in cases where little data is available. A prototypical example of this is the one-shot learning setting, in which we must correctly make predictions given only a single example of each new class. In this project , we explore a method for learning siamese neural networks which employ a unique structure to naturally rank similarity between inputs. Once a network has been tuned, we can then capitalize on powerful discriminative features to generalize the predictive power of the network not just to new data, but to entirely new classes from unknown distributions.
Learning effective features for machine learning tasks can be computationally demanding, particularly when dealing with limited data scenarios. Performing accurate predictions based on a single example per class is a challenging scenario for one-shot learning. This paper explores the use of Siamese neural networks for addressing one-shot image recognition tasks. Siamese networks are structured to learn discriminative features by comparing pairs of input images, allowing for the assessment of similarity and dissimilarity between them. Utilizing a convolutional architecture, our approach achieves state-of-the-art performance in one-shot image recognition, surpassing other deep learning models. This research contributes to advancing the field of one-shot learning methodologies and underscores the importance of leveraging Siamese networks for feature extraction and similarity assessment in the context of image recognition with minimal training data.
The Siamese network design comprises two identical subnetworks, each processing one of the inputs. Initially, the inputs undergo processing through a convolutional neural network (CNN), which extracts significant features from the provided images. These subnetworks then generate encoded outputs, often through a fully connected layer, resulting in a condensed representation of the input data.
The CNN consists of two branches and a shared feature extraction component, composed of layers for convolution, batch normalization, and ReLU activation, followed by max pooling and dropout layers. The final segment involves the FC layer, which maps the extracted features to the ultimate classification outcomes. A function delineates a linear layer followed by a sequence of ReLU activations and a series of consecutive operations (convolution, batch normalization, ReLU activation, max pooling, and dropout). The forward function guides the inputs through both branches of the network.
The Differencing layer serves to identify similarities between inputs and amplify distinctions among dissimilar pairs, accomplished using the Euclidean Distance function: Distance(x₁, x₂) = ∥f(x₁) – f(x₂)∥₂ .
Two types of loss is used :
1.The mathematical equation for Mean Absolute Error (MAE) or L1 Loss is: MAE = (1/n) * Σ|yᵢ - ȳ|
2.The mathematical equation for Binary Cross-Entropy Loss, also known as Log Loss, is: L(y, f(x)) = -[y * log(f(x)) + (1 - y) * log(1 - f(x))]
This property enables the network to acquire effective data representations apply that to fresh, unseen samples. Consequently, the network generates an encoding, often represented as a similarity score, that aids in-class differentiation.
Depict the network’s architecture in the accompanying figure. Notably, this network operates as a one-shot classifier, negating the need for many examples per class.
Here’s how they typically work:
1.Architecture: Siamese networks consist of two identical subnetworks (often referred to as Siamese twins) that share the same architecture and weights. 2.Training: During training, pairs of images are passed through the network. The network learns to produce similar feature representations for images of the same class and dissimilar representations for images of different classes.
3.Loss Function: A contrastive loss function is commonly used to train Siamese networks. This loss function encourages the network to minimize the distance between feature representations of similar images and maximize the distance between representations of dissimilar images.
4.One-shot Learning: Once trained, the Siamese network can recognize whether a pair of images belong to the same class or not, even if it has only
seen one or very few examples of each class during training.
In conclusion, the implementation of a One Shot Face-Recognition system utilizing Siamese Network architecture offers a promising solution to the challenges of face recognition tasks, particularly in scenarios with limited training data. By leveraging the Siamese Network's ability to learn subtle facial features and encode facial similarities, this approach demonstrates remarkable accuracy even with a single reference image per individual. Furthermore, its versatility extends to various applications, including security systems, access control, and personalized user experiences. As advancements continue in deep learning and computer vision, the integration of Siamese Networks into face recognition technology holds great potential for enhancing efficiency and reliability in real-world applications , ranging from security and authentication to personalization and healthcare. However, their deployment must be accompanied by robust regulations, ethical considerations, and privacy safeguards to ensure responsible and fair use.
Aniket K
Niranjan
Vinit K
Anvitha
Kenisha
Radhika
Sahana
Meet Link: https://meet.google.com/kwt-dhdx-ybi
Report prepared on May 8, 2024, 6:54 p.m. by:
Report reviewed and approved by Aditya Pandia [CompSoc] on None.