Srinivasa Manohar Karlapalem, Senior Software Engineer, Open Source Technology Center, Software and Services Group Deep learning is increasingly becoming popular for visual understanding use-cases on edge devices such as image classification, object detection etc. There has been an increasing demand for running basic computer vision and deep learning models on edge devices due to factors such as privacy, security etc. The information from these models is then sent to cloud for further processing. We demonstrate object detection using Single-shot multi-box detection (SSD) running on edge devices. There are several implementations of SSD with deep neural nets such as Googlenet, VGG etc. which provide low-throughput object detection solutions on edge devices. We demonstrate a new SSD network with Squeezenet for high throughput object detection on edge devices with accelerators using Intel’s Computer Vision SDK. We will also demonstrate how to easily deploy inference solutions on edge devices using AWS Greengrass lambdas.