ML-Based Object Recognition Device is a research project researching the feasibility of implementing machine learning (ML) based object recognition on a field programmable gate array (FPGA). This research will aid designers in creating a product that can be used by the public in order to harness the power of AI. We will also research the potential improvement in the speed and number of parameters used after adding a tensor contraction layer to a convolutional neural network. Key technologies used in this system include C++ simulations run on a PC that will allow us to measure the potential benefits of adding a tensor contraction layer to a convolutional neural network. We performed experiments on the MNIST dataset which is a collection of handwritten digits and we measured the accuracy at which the model could classify the digits. We found that the tensor contraction layer model uses half of the parameters but it consumes 10x more time processing a single image.