MNIST digits image similarity search by Indexing with Annoy and using trained embeddings from a Siamese Net with Triplet Loss .

Image quality assessment using siamese convolutional neural network architecture

This work proposes a feature refined end-to-end object tracking framework with a balanced performance using a high-level feature refine tracking framework. The feature refine module enhances the target feature representation power that allows the network to capture salient information to locate the target.

I create a Neural Network model using Siamese architecture to distinguish between real user and impostor when inputting pin number on various mobile app payment method.