The Doppler Quarterly Summer 2018 - Page 17

Inception-v3, which is a deep convolutional neural network. It was trained for the ImageNet Large Visual Recognition Challenge using data from 2012, with 1000 classes of images, such as "Zebra", "Dalmatian", and "Dishwasher." From this, we were able to create specific classifications for smoke, fire and gauge readings in our demo. Creating a model from scratch can take months, even with highly skilled analysts working together. With transfer learning, a new model can be built in hours, often with only a few dozen images to start, and the skills needed for software development in program- ming languages such as Python or C++. locally, to detect smoke, fire, and the state of the ana- log gauge (green, yellow, or red). There are two output streams from the real-time edge processing of the video analytics (Figure 1): 1. A real-time dashboard showing the current state and timeline for the processing. This would be monitored by someone on the factory floor, and also sends alerts when problems are detected. 2. A near real-time stream of alerts sent through an IoT gateway to GCP via Cloud IoT Core over MQTT. These alerts are pushed to a dashboard hosted in the cloud, as well as Stackdriver Monitoring, where further threshold-based alerting can be performed. Utilizing public cloud resources is the perfect fit for training models, as the processing of image sets can be distributed to parallel process- ing resources such as GPUs or STATUS TPUs 2 , which are released when processing is complete. We used FIRE the Google Cloud Platform (GCP) and Cloud ML Engine to build and train the graph, which was then SMOKE exported and deployed to run at the edge, which in this case is a RPM simulated factory floor. Cloud ML Engine utilizes an open source machine learning technology called TensorFlow (also developed by Google). We ran the same Ten- sorFlow Serving technology for the Anomalous Alert Timeline edge processing of the video stream. Using a sampling rate of 3 frames per second in a video stream from an IP camera pointed at the demo box, each frame is sent to the Ten- sorFlow Serving processor running Acknowledge Acknowledge Acknowledge All In Progress Resolved HMI Danger Detected on Video 4/9/2018 3:09:50 PM EDT Smoke Detected on Video 4/9/2018 3:09:50 PM EDT Fire Detected on Video 4/9/2018 3:09:50 PM EDT Assign Details Acknowledge Assign Details Acknowledge Assign Details Acknowledge Figure 1: Fire, Smoke and RPM Alerting Dashboard SUMMER 2018 | THE DOPPLER | 15