Real-Time Applicability Analysis of Lightweight Models on Jetson Nano Using TensorFlow-Lite

Deep learning models have recently acquired prominence due to their adaptability to constrained devices. Because of this possibility, a significant number of studies in the fields of IoT and Robotics are being done with the goal of deploying deep learning models on resource-constrained applications. A variety of lightweight models are now available that can perform computer vision tasks on constrained devices including the Jetson Nano. However, several enhancements are still needed if this field of research has to prosper in the future. This study was carried out with the aim of comparing and contrasting the lightweight models provided by TensorFlow in order to assess them and ascertain how close they are to practical reality. The conclusions not only present the observed outcomes but also provide insight into the models, attempting to identify potential improvements.