Intel on AI

In this Intel on AI podcast episode: Safety and productivity issues are crucial for engineering, manufacturing, and industrial organizations. Ensuring that workers are in their safe and respective area not only helps prevent injuries but also increases productivity and compliance for a company. Apoorva Ankad, Head of Computer Vision and AI Group at Deevia Software, joins the Intel on AI podcast to discuss Deevia’s AI powered monitoring system which supports several different industries with AI powered vision solutions. He highlights how Deevia’s solution can leverage an organization’s existing CCTV (closed circuit television) camera infrastructure and existing Intel-based infrastructure to accomplish their customer’s training and inferencing needs. He talks about how Deevia’s solution has been used to help detect and analyze factory worker’s ergonomic positions to help analyze if workers are conducting their tasks in a safe manner. This can help maintain worker safety by sensing when they are not operating in a safe way and alerting the worker that they need to adjust their work to a more ergonomically correct action. Apoorva also discusses how Deevia has pivoted their technology to help create a social distancing monitoring system where camera sensors can detect if social distancing is correctly happening in a public area. Where a lapse is detected, an audio message can be delivered via speakers as a reminder to adhere to distancing protocols and help keep the general public safe by helping to remind everyone to stay socially distanced. To learn more, visit: deevia.pw Visit Intel AI Builders at: builders.intel.com/ai

21 Juli 202014min

In this Intel on AI podcast episode: Industrial operations such as metal manufacturing need to keep track of how well their facilities, time, and materials are being utilized in order to be as productive and profitable as possible. Yet, there can be 100s of sensors in a manufacturing plant and capturing, analyzing, and making predictions based on all that data can be very difficult. Also, ensuring the proper working or equipment can also ensure the safety of industrial workers and minimize wasted materials. Hitesh Mittal, the Director Business Development from Tvarit GmbH, joins the Intel on AI podcast to discuss how Tvarit is working to optimize the business processes of their customers. He described a specific use case where Tvarit worked to optimize operations of a steel manufacturing plant using supervised machine learning algorithms to predict the health of mechanical components. Analyzing and predicting equipment utilization reduces waste & downtime, increases safety and profitability and Hitesh describes how the training and inferencing were done on Intel Xeon Scalable processors. Hitesh also emphasizes how Tvarit works closely with the Intel AI Builders program to optimize their solution for Intel technology and praised the program for the amount of help Tvarit received from the Intel team. To learn more, visit: tvarit.com Visit Intel AI Builders at: builders.intel.com/ai

10 Juli 202012min

In this Intel on AI podcast episode: The COVID-19 coronavirus, since its initial outbreak, has quickly become a global pandemic and the inadequate amount of lab tests available for people suspected of infection have posed serious risks to public health and efforts in containing the virus. Jingwen Jia (Wendy), the Assistant General Manager at Huiying Medical (HYHY), joins the Intel on AI podcast to discuss how the HYHY imaging diagnostic solution assists healthcare providers to detect and diagnose potential COVID-19 infections by analyzing computed tomography (CT) chest scans with AI-powered algorithms. She discusses how the HYHY solution is a complementary tool to help doctors make diagnosis quickly by analyzing the ground-glass opacity (GGO) and other indicators revealed in lung CT imagery. The HYHY solution has already been deployed in 30+ hospitals throughout China. Wendy emphasizes how, with the help of AI, the HYHY solution can help doctors detect the virus quickly, helping patients receive the care they need faster and assisting with the tracking and containment of the COVID-19 virus. She also discusses how HYHY has collaborated with Intel through the Intel Capital and Intel AI Builders programs to optimize their solutions to run on Intel architecture and the Intel Distribution of OpenVINO toolkit to help give healthcare providers a powerful tool to help combat the COVID-19 pandemic around the world. To learn more, visit: en.huiyihuiying.com Visit Intel AI Builders at: builders.intel.com/ai

6 Juli 202014min

In this Intel on AI podcast episode: Retail shoppers around the world are always looking to have a more personalized, convenient, and overall better shopping experience when they visit stores. With the advent of advanced cameras, sensors, and AI technology, smart stores are eliminating the need for customers to wait in line, scan their purchases, or even complete a payment transaction with a cashier. Also, during the worldwide pandemic it can be difficult for retail staff to limit their exposure to potentially infected people throughout the course of their workday. For consumers as well, maintaining safe distances can be nearly impossible when at a busy store or when interacting directly with a store cashier. Mark Perry, the Global Business Director at Cloudpick, joins the Intel on AI podcast to talk about how Cloudpick’s AI-powered smart store solution is providing customers with enhanced shopping experiences while also working to help keep them safe. He describes how Cloudpick’s system can automatically recognize a customer when they enter the store as well as the products that the customer gathers and automatically charge costs to the customer’s account without having to interact with a cashier or scan their items. Mark also talks about how during the Covid-19 pandemic, retail stores being able to limit staff interactions with customers and allowing customers to avoid touching checkout machines helps customers staff avoid potential exposure. He also discusses how optimizing the Cloudpick solution for Intel architecture and using the Intel distribution of OpenVINO has helped enable Cloudpick’s solution to operate fater and provide consumers with better, safer shopping experiences. To learn more, visit: cloudpick.com Visit Intel AI Builders at: builders.intel.com/ai

24 Juni 202016min

In this Intel on AI podcast episode: As wind turbines proliferate and grow in size and complexity, the biggest challenge to Wind Energy is the high cost of Operations and Maintenance. Manual inspection and maintenance is dangerous and expensive. With the advent of drones, gathering maintenance footage has become much easier, but without the use of AI technology, inspecting tons of footage and data is time consuming, expensive, ineffective. One defect can potentially incapacitate an entire turbine, however automation of maintenance can significantly improve the value and cost of wind energy. Alberto Gutierrez Ph.D., Chief Data Scientist at HCL America, joins the Intel on AI podcast to talk about HCL’s deep learning (DL) based wind turbine defect detection solution and how it is helping to drive sustainable energy today. He illustrates how HCL’s solution enables wind energy operators to utilize drone technology to capture images of turbines and use deep neural network (DNN) computer vision algorithm to find potential defects in those turbines. Some of the defects that are often detected include visible defects on blade surfaces like missing teeth in VG (Vortex Generator) or panel and blade edge corrosion. Alberto describes how using AI and drones to address this workload is ultimately a safer and less expensive option that helps make wind energy cheaper and more attractive as an alternative, clean energy source. He also discusses how HCL has collaborated closely with the Intel AI Builders program to optimize their solution’s DL model using the Intel Distribution of OpenVINO toolkit to process video stream, image segmentation and object detection. To learn more, visit: builders.intel.com/ai/membership/hcl Visit Intel AI Builders at: builders.intel.com/ai

17 Juni 202013min

In this Intel on AI podcast episode: 4G/ LTE network is a preferred method of information transfer today and is becoming more and more crucial in our extremely connected lives. To keep up with the ever-increasing volume of traffic, the network is constantly changing and becoming more and more complex. Legacy rules-based network automation techniques are not working, and communication service providers need to use predictive health analytics to monitor, predict and optimizing the behavior of 4G/LTE network continuously. Networks need to become ‘intelligent’ and can take care of themselves?. Subhankar Pal, the Assistant Vice President of Research and Innovation at Altran, joins the Intel on AI podcast to discuss how Altran’s NetAnticipate framework is driving state-of-the-art self-learning networks through continuous self-feedback. Subhankar illustrates how Altran’s NetAnticipate uses advanced deep learning (DL) models for channel quality prediction and health analytics of 4G/LTE radio networks. He talks about how the solution involves network behavior prediction using key performance indicators in multi-cell mobility scenarios along with several regression and classification models chained together to achieve its network prediction. Subhankar also describes how the Intel AI Builders team helped with optimization testing of Intel optimized Python and Tensorflow to enable Altran to reduce training time and improve model performance so their customers can use existing Intel based hardware to achieve their network automation. Finally, Subhankar discusses the future of 5G technology and how Altran is enabling the future of LTE networks. To learn more, visit: altran.com Visit Intel AI Builders at: builders.intel.com/ai

17 Juni 202016min

In this Intel on AI podcast episode: Sparse modeling methods can improve the interpretability of predictive AI models and is widely used in academia today. Yet despite advances in the field, issues remain when sparse modeling meets real-life applications. Adrian Sossna, the Chief Marketing Officer at HACARUS, joins the Intel on AI podcast to talk about how the SPECTRO visual software inspection module can make sparse modeling more available. He highlights how SPECTRO enables factory automation by vastly reducing the amount of reclassification needed by human inspectors. Adrian talks about how this enables AI models to be trained faster with less data while achieving accurate results specifically targeted for production of precision parts, metals, plastics and other materials. SPECTRO contains explainability features that allow detection of defects within manufacturing and provides businesses to make improvements in their processes because they have visibility into the detection made by the software. Adrian also talks about how working with the Intel AI Builders program has allowed HACARUS to run SPECTRO on Intel Optimized Python and achieve impressive performance improvements and has been very powerful for HACARUS to deliver a better experience to their customers. To learn more, visit: hacarus.com Visit Intel AI Builders at: builders.intel.com/ai

2 Juni 202015min

In this Intel on AI podcast episode: In industrial and manufacturing settings plant safety issues like oil spills, chemical spills, and PPE (personal protective equipment) violations can happen often. These issues put lives at risk, the environment in danger, and bring down productivity and profitability of the organization. Yet, addressing these issues automatically with AI is a challenging potential. The effort to capture, create and analyze a data set for image and video annotation is immense. Sajin Payandath, the Lead Data Scientist at Flutura, joins the Intel on AI podcast to illustrate how Flutura’s CerebraVision solution addresses these issues of safety and security specific to oil & gas, manufacturing, and heavy engineering sectors. He describes how CerebraVision is a central safety monitoring system that uses AI analysis of CCTV video feed to detect safety violations occurring within an industrial plant environment. This allows organizations to detect and respond to safety issues in real-time. The solution can also potentially be used to improve the productivity by combining visual intelligence with sensor data to analyze and improve industrial processes. Sajin talks about how Flutura has been working to adapt their solution to help organizations during the Covid-19 pandemic to detect and alert social distancing violations within the workplace. He also Highlights how Flutura has worked with Intel to optimize Flutura’s training and inference workloads to ensure they run efficiently and with increased accuracy and performance on Intel architecture. To learn more, visit: flutura.comflutura.com Visit Intel AI Builders at: builders.intel.com/ai

21 Maj 202012min