{"id":458,"date":"2021-06-07T09:19:12","date_gmt":"2021-06-07T16:19:12","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/?p=458"},"modified":"2021-06-11T11:28:49","modified_gmt":"2021-06-11T18:28:49","slug":"case-study-davidovich-group-keeps-its-eyes-on-inventory-with-azure-3d-and-ai-in-the-cloud","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/case-study-davidovich-group-keeps-its-eyes-on-inventory-with-azure-3d-and-ai-in-the-cloud\/","title":{"rendered":"Case Study: Davidovich Group keeps its eyes on inventory with Azure 3D and AI in the cloud."},"content":{"rendered":"

Overview\u00a0<\/span><\/b>\u00a0<\/span><\/p>\n

Davidovich\u00a0Group\u00a0is\u00a0a\u00a0producer\u00a0and wholesaler\u00a0of bread and pastry products\u00a0that\u00a0has\u00a0been supplying Israelis\u00a0with baked goods\u00a0for over 50 years.\u00a0A successful bakery strives to meet the fluctuating demand for their products. The bakery operations involve\u00a0creative offerings,\u00a0but also\u00a0product\u00a0and a repeatable process. It’s a regulated industry\u00a0where core products\u2019 price is controlled leaving low profit margins.\u00a0The raw materials and the products have limited shelf life that needs to be monitored and managed. Profitability relies on an efficient, automated operation, run at scale.\u00a0Any disruption to production or shrinkage has an out-sized effect on the bottom line. Davidovich adopts modern technology and methods to help them improve efficiency and reduce costs.<\/span>\u00a0<\/span><\/p>\n

 <\/p>\n

Situation<\/span><\/b>\u00a0<\/span><\/p>\n

One of the\u00a0perennial\u00a0problems\u00a0logistics operation\u00a0faces\u00a0is\u00a0product\u00a0shrinkage\u00a0and\u00a0addressing it head-on\u00a0is a business priority.\u00a0With\u00a0good inventory management practices\u00a0food providers can\u00a0optimize stock and reduce unnecessary loss\u00a0due to spoilage,\u00a0theft,\u00a0and\u00a0operational errors.\u00a0In\u00a0Davidovich\u2019s case,\u00a0they have\u00a0automated portions of their warehouse operations\u00a0and\u00a0practice\u00a0best-in-class inventory management\u00a0and still they\u00a0recognize\u00a0that staying on top of technology\u00a0is the key to remaining competitive in a commodity market.\u00a0<\/span>\u00a0<\/span>As\u00a0part of their\u00a0digital\u00a0transformation,\u00a0the company\u00a0is\u00a0updating systems and operations\u00a0to reduce loss, improve delivery times and\u00a0increase\u00a0order accuracy.\u00a0<\/span>\u00a0<\/span><\/p>\n

Davidovich researched solutions and reached out to\u00a0potential\u00a0providers that\u00a0would help them\u00a0update their systems to capture, identify and measure palleted\u00a0products.\u00a0The\u00a0requirements\u00a0were clear\u2014any solution\u00a0must\u00a0integrate with\u00a0the\u00a0existing\u00a0infrastructure\u00a0and\u00a0could not disrupt production.<\/span>\u00a0<\/span><\/p>\n

 <\/p>\n

Challenges<\/span><\/b>\u00a0<\/span><\/p>\n

Davidovich\u00a0explored existing\u00a0market\u00a0solutions,\u00a0but they\u00a0did\u00a0not\u00a0satisfy the requirements\u00a0due to\u00a0a\u00a0large upfront investment\u00a0and\u00a0ongoing\u00a0maintenance costs.\u00a0The warehouses would\u00a0also\u00a0need to be retro-fitted\u00a0that would require operations\u00a0be\u00a0halted for some time.\u00a0That\u2019s when\u00a0Davidovich\u00a0turned to Agent Factory,\u00a0experts in\u00a0applying\u00a0custom AI solutions\u00a0for\u00a03D\u00a0machine vision.\u00a0What the bakery needed was a\u00a0means to accurately determine what product they had on hand as the starting point to\u00a0more robust inventory management\u2014a solution that would\u00a0literally identify product coming and going from their warehous<\/span>es\u00a024 hours a day, every day.<\/span>\u00a0<\/span><\/p>\n

 <\/p>\n

Solution<\/span><\/b>\u00a0<\/span><\/p>\n

There were\u00a0multiple technical challenges\u00a0to keep the\u00a0chosen\u00a0solution\u00a0simple to maintain and within budget.\u00a0Traditional\u00a02D cameras\u00a0do not have the notion of depth and without\u00a0it\u00a0dimensioning cannot work.\u00a0Laser-based systems or\u00a0multiple\u00a0camera arrays\u00a0could potentially capture\u00a03D\u00a0depth\u00a0images\u00a0but require sophisticated equipment which the require cargo to\u00a0be placed at a designated location to\u00a0capture dimensions.\u00a0<\/span>\u00a0<\/span><\/p>\n

\"Image<\/a>
Figure 1: Multiple cameras, each with a small compute device connected to a single PC.<\/em><\/span><\/figcaption><\/figure><\/p>\n

 <\/p>\n

The solution\u00a0Davidovich\u00a0ultimately chose is based on Microsoft\u2019s 3D Time-of-Flight (ToF) technology that can capture high-quality, precise 3D data of objects in motion. ToF-based cameras positioned at the entrances and exits of the company\u2019s warehouses capture images of incoming and outgoing freight including and algorithmic processing determines the dimensionality.\u00a0The solution aligned\u00a0seamlessly with\u00a0Davidovich\u2019s\u00a0existing\u00a0processes,\u00a0and camera\u00a0installation\u00a0only\u00a0required\u00a0simple\u00a0rack\u00a0mounts\u00a0and\u00a0subsequent\u00a0calibration\u00a0to\u00a0ensure correct positioning.<\/span>\u00a0<\/span><\/p>\n

\u201cLeveraging Microsoft\u2019s 3D Time of Flight technology and the Azure cloud platform we were able to develop a solution rapidly and cost-effectively with the accuracy, field of view and range required.\u201d<\/span><\/p>\n

Boaz Ein-Gil,<\/strong> CEO, Agent Factory<\/p><\/blockquote>\n

The solution also had to support high data rates from multiple capturing points at all the entrances of a storage area and track both incoming and outgoing freight. The amount of data captured is large and complex to process; a single depth camera produces ~2.5TB (at 15fps) of unfiltered data a day. With so much data to store and process the solution had to balance the costs of processing data on-prem with processing in the cloud. The data is processed on prem so only the most useful data is sent to the cloud for analysis to avoid unnecessary bandwidth costs.<\/span>\u00a0<\/span><\/p>\n

\u201cIntegrating our cloud analysis with Azure Cognitive Services allowed us to bring world-class label detection into our solution in no time.\u201d<\/span><\/p>\n

Avihai Lifschitz,<\/strong> Chief Scientist,\u00a0Agent Factory<\/p><\/blockquote>\n

Agent Factory chose Azure IoT Edge as the runtime platform to run process-intensive workloads on top of Jetson Nano compute with Microsoft-based 3D ToF cameras. Azure IoT Edge is also used as the gateway to Azure and offers a ready-made infrastructure to support remote deployment and management and monitoring of the on-prem software components which saved considerable development time.<\/p>\n

 <\/p>\n

 <\/p>\n

\"Image<\/a>
Figure 2- Cameras positioned at entrances and exits capture 3D data. Data is processed at the edge and optimized before sending it to the cloud.<\/em><\/span><\/figcaption><\/figure><\/p>\n

 <\/p>\n

In addition, Agent Factory designed a resilient, high-performance processing pipeline to drastically reduce data size down to ~0.01% of total captured data so that only the most useful frames from each activity are uploaded to the cloud. The pipeline includes a dedicated, proprietary trained DNN classifier that uses depth-based images as input.<\/p>\n

On Azure, combining Blob storage, Azure Functions and Microsoft Cognitive Services<\/a> allows for cost efficient processing of uploaded frames to extract label data and dimensions of the freight. In addition, Azure App Service allowed Agent Factory to rapidly develop an online portal that makes the optimized data available to Davidovich through a secure web portal.<\/p>\n

 <\/p>\n

\"Image<\/a>
Figure 3- Azure services in use for this solution<\/em><\/span><\/figcaption><\/figure><\/p>\n

 <\/p>\n

Benefits<\/strong><\/p>\n

Leveraging a selection of proven combination of Azure technologies and services dramatically saves development costs. The processing of 3D data and images requires complex algorithms with high computation demands that are more easily met in the cloud than with on-prem processing. With that, providing seamless experience requires constantly capturing 3D data that quickly becomes too large to be pushed to the cloud requiring a hybrid of on-prem and cloud components.<\/p>\n

 <\/p>\n

“With this solution, we are getting the desired extra level of assurance managing our inventory with a minimal investment.”<\/span><\/p>\n

Benny Davidovich, <\/strong>Chief Technology Officer,\u00a0Davidovich Group<\/p><\/blockquote>\n

 <\/p>\n

The 3D cameras featuring Microsoft\u2019s best-in-class breed, ToF depth sensing technology provide the wide fields of view and image quality necessary to capture freight moving in and out of the warehouse. Agent Factory designed the solution so that the on-prem hardware and software that captures activities 24\/7 can distill the captured data to a reduced footprint that can be easily pushed to the cloud. Once data reaches the cloud, it is processed to extract the business value by executing a dimensioning algorithm and combines it with label data, to produce digital logs with visual evidence of all freight movement.<\/p>\n

Developers considering 3D machine vision AI solutions have the advantage building solutions with Azure IoT Edge on-prem and Azure cloud services because they serve as the foundation to cost-efficient 3D solutions. Solutions based on the Azure Depth Platform readily scale based on customer need and can reach hundreds of deployed cameras working in parallel. Customers can make their logistic processes smarter which directly translates to cost savings. Davidovich saw a return on their investment within two quarters.<\/p>\n","protected":false},"excerpt":{"rendered":"

\u00a0Davidovich\u00a0chooses Microsoft\u2019s 3D Time-of-Flight (ToF) technology to capture high-quality, precise 3D data of objects in motion in the company’s warehouse to improve inventory management. <\/p>\n","protected":false},"author":63233,"featured_media":461,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1,18,19],"tags":[],"class_list":["post-458","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-azure-depth-platform","category-robotics","category-sensors"],"acf":[],"blog_post_summary":"

\u00a0Davidovich\u00a0chooses Microsoft\u2019s 3D Time-of-Flight (ToF) technology to capture high-quality, precise 3D data of objects in motion in the company’s warehouse to improve inventory management. <\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/posts\/458","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/users\/63233"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/comments?post=458"}],"version-history":[{"count":0,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/posts\/458\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/media\/461"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/media?parent=458"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/categories?post=458"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/azure-depth-platform\/wp-json\/wp\/v2\/tags?post=458"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}