Shelf Image Recognition and Items Counter Dashboard POC

$8-15 USD / hour

Posted 1 minute ago

•

Ends in 6 days

$8-15 USD / hour

Design: Build a Proof of Concept (POC) as a dashboard that allows a BM to select their customer from a drop down, select a category and subcategory, and then be able to upload an image of store shelf where you are able to provide a list of products, brand, package types, prices, and description when available. The data from image should show up in a table format in dashboard to know the number of times specific Store Shelf Items in an uploaded image appears on the Shelf Image provided. The app, would be like a form-based app where the BM and Product Information would be made available on the screen, but all the BM, Customer and Product Data shall be pulled from Tableau. The App UI should have all the BM info ( Name, BM User_ID), Customer Info(Customer ID, Customer Name, Customer Location), Product Info (Brand, Packet Type, Price, Item Description, Weight). On the App UI, there should be a capability of capturing a picture. The contents of this picture should be understood and translated into the Products Info as describe above. The data translation process should be as thus: first, move all the data (BM, Customer & Product Data, and uploaded Image) into Azure BLOB. Then the data is passed from the BLOB into the Microsoft Foundry so that the LLM can easily translate the uploaded image and passed the translate data of the image along with other Product details, plus the BM's data into an Azure SQL database. no actual URL available as of now build the mock of this task and assume some fields for user creation, also run locally just a demo. Please provide the output of the Products' data deduced from the image along with the Customer and BM data in an organized format in an artifact. Use the configured Tableau MCP and other MCPs as needed for your datasources

Project ID: 40811738

About the project

23 proposals

Open for bidding

Remote project

Active 3 secs ago

Place your bid

Bid amount Email address

Benefits of bidding on Freelancer

Set your budget and timeframe

It's free to sign up and bid on jobs

23 freelancers are bidding on average $16 USD/hour for this job

User Avatar
Flag of PAKISTAN

I can build a locally running POC that lets a BM select a customer and category, capture or upload a shelf photo, and review the detected products alongside BM and customer details. I’ll create a form-based dashboard with a product table showing brand, package type, price, description, weight, and the number of times each item appears. I’ll use the configured Tableau MCP for available data and a clearly labeled mock dataset where live access is unavailable. The workflow will stage the records and image in Azure Blob Storage, pass the image to Microsoft Foundry for analysis, match results against the product catalog, and save the structured output to Azure SQL. I’ll also provide an organized JSON or CSV artifact for the demo. Fields the image cannot reliably establish, especially prices, will come from product data or be marked unavailable rather than guessed. I’ve built data-driven dashboards and Azure-backed workflows, and I’ll keep the Tableau and AI connections easy to replace when your production access is ready. Do you have a sample shelf image and product catalog I should use to validate the demo?

User Avatar
Flag of MALAYSIA

Image recognition for shelf counts usually comes down to clean training data more than the model itself. For a POC I would build a simple web dashboard where the BM picks a customer, uploads shelf photos, and gets item counts back using Azure Custom Vision, with results pushed into a Tableau view. I can start today and have a working demo in 3 days. Budget and timeline here are rough, based on the post, and may shift once we talk through the real data setup. Want me to send a quick scope doc?

User Avatar
Flag of TURKEY

Hi there, I noticed you're looking to build a dashboard that integrates image recognition capabilities to list store shelf products. I can leverage my experience in Azure, data management, and web application development to create a proof of concept that meets your requirements. For the image processing, I will develop a mechanism that captures the uploaded image and utilizes Azure BLOB for data storage. The data will then be processed through Microsoft Foundry to extract product details, including brand and packaging type, which will be stored in an Azure SQL database. Ensuring that this workflow integrates smoothly with Tableau for data visualization will also be a priority. I can deliver a working mock-up that fulfills these requirements within a week. Regards, Ali

User Avatar
Flag of INDONESIA

IF YOU ARE NOT HAPPY YOU DON'T PAY! Vision LLMs read brands and price tags well but often miscount on a crowded shelf. So I'd have the model return one row per facing and add them up in code instead of asking it for a total. I'd mock Tableau, Blob (Azurite) and Azure SQL locally behind the same interfaces, so switching to the real endpoints later is just config. You'd get the BM/customer/category form, camera capture or upload, a product table with counts, and the combined output artifact. Do you have a sample shelf photo and a Foundry vision deployment, or should the demo use a stand-in? Send me a message and I'll explain exactly how I'd approach your project.

User Avatar
Flag of UNITED KINGDOM

You need a dashboard where BMs can upload shelf images and automatically extract product data, brand, price, package type, description, structured in a table alongside customer and BM info from Tableau. I'll build a mock POC that flows image → Azure BLOB → Microsoft Foundry LLM for recognition → Azure SQL for storage, then render results on the dashboard. The frontend will include dropdown selectors for customer and category, an image upload (with live capture option), and a results table showing detected products with their details. Since this is a local demo without live Tableau/Azure endpoints, I'll mock Tableau data pulls and simulate the LLM translation step so you can see the full flow end-to-end. The app will be a React dashboard with form-based layout displaying BM info, customer details, and an image upload interface. Once an image is processed (simulated LLM output), the detected products populate a table. I'll structure everything modular so when you integrate real Azure services and Tableau APIs later, the scaffolding is already there. Regards, Savi

User Avatar
Flag of PAKISTAN

Hi, You need a Proof of Concept (POC) dashboard that allows Business Managers to upload images of store shelves and extract product information efficiently. What specific image recognition model do you plan to use for translating the uploaded images? Here's how I'd do it: - Set up the front-end app with dropdowns for customer selection and file upload functionality. - Integrate Azure BLOB for image storage and pass data to Microsoft Foundry for processing. - Organize the extracted product information and relevant metadata into a structured format displayed in a table within the dashboard. I can have a working version of the core piece ready on a test link within a day. What format do you prefer for the organized product data artifact? Toqeer

User Avatar
Flag of PAKISTAN

Hi, I understand you need a local dashboard POC where a BM selects customer, category, and subcategory, uploads or captures a shelf image, and receives organized product details and item counts. I can build the mock flow with assumed user fields, Tableau-style source data, Azure Blob storage, Microsoft Foundry image processing, and Azure SQL output. The dashboard can show BM, customer, brand, package type, price, weight, description, and detected quantities in a clear table, with an artifact containing the final BM, customer, and product data. I will keep the demo easy to run locally and structure it so real Tableau and Azure connections can be added later. For the POC, should the shelf image output use a fixed product catalog for matching, or should it identify unknown products as well? Please start the chat and let’s discuss the image fields, expected product format, and preferred local dashboard framework.

User Avatar
Flag of INDONESIA

a shelf photo turning into a table of brands, package types and item counts for your bm dashboard is very doable as a local demo, even with the tableau and azure pieces mocked my team built a store front with product and order management, so a form and table dashboard like this is familiar ground i'll stay on it and sort out any bugs until the demo runs the way you want do you have a sample shelf photo and a tableau data extract i can mock the demo against? happy to get started once i hear back Sela

User Avatar
Flag of UKRAINE

Hello, This POC needs more than image labeling: the important part is connecting a shelf photo to BM, customer, and product context while preserving the result in an organized artifact. I can build a local demo dashboard with assumed user fields, customer/category selectors, image upload or camera capture, and a product-count table. I’ll mock the Tableau source through the configured MCP, stage the image and metadata in an Azure Blob-style layer, then demonstrate the Microsoft Foundry vision/LLM translation into Azure SQL. The attached shelf image can be used as the demo input after you confirm the intended product fields. I can deliver the first working local milestone within one week, including setup notes and sample output. Best regards, Bohdan

User Avatar
Flag of UKRAINE

Hi there, What Tableau fields are available for BM, customer, and product lookup, and should the local demo mock those schemas exactly? How should the shelf image output be validated: object detection counts only, or also brand, pack type, price, and description mapping when visible? The main challenge is combining form data, image analysis, and a clean demo flow without a live backend. A practical approach is a local dashboard that loads mocked Tableau-like data, accepts an uploaded shelf image, stores raw inputs in Azure BLOB placeholders, then sends the image and metadata into a Foundry-style translation step before writing the structured result into Azure SQL. A key risk is inaccurate item recognition from one shelf photo, especially when packaging is partially hidden or labels are small. That can be handled by keeping the POC transparent: show confidence, allow manual correction, and separate detected items from inferred product fields. For the demo, the UI can present BM, customer, and product selection, camera upload, and an organized artifact with the extracted rows and counts. Test coverage should include empty selections, duplicate products, and image edge cases, then run locally as a self-contained proof of concept. Best, Hlib

User Avatar
Flag of UNITED STATES

Dear Client, Thank you for sharing the opportunity to collaborate on your Shelf Image Recognition and Items Counter Dashboard POC. I am excited about the innovative approach you are taking to streamline shelf auditing and product data management, and I would be delighted to contribute my expertise to bring your vision to life. Having delivered successful Proof of Concept dashboards and AI-powered data solutions for retail and FMCG clients, I understand the importance of seamless data integration, intuitive UI/UX, and accurate image-to-data translation. Your requirement to connect BM, Customer, and Product data from Tableau MCP, enable image uploads, and leverage Azure and Microsoft Foundry for intelligent image recognition aligns perfectly with my skill set. My background in Azure Blob Storage, Tableau integration, and web application development ensures I can architect a robust, scalable, and demo-ready solution. To address your needs, I propose building a local, form-based dashboard where BMs can select their customer, category, and subcategory, and upload shelf images. The app will simulate pulling BM and Customer data from Tableau, and upon image upload, will process the image to extract product details using a mock LLM flow. All data—including the original and translated product information—will be organized and displayed in clear, searchable tables. The data flow will be designed to mimic the transfer from Azure Blob to Foundry, and subsequently to Azure SQL for storage and further analysis. I am enthusiastic about delivering a compelling demo that demonstrates the full workflow, while ensuring all data artifacts are organized and accessible for your review. I look forward to discussing how my experience can help you achieve your goals for this POC. Best regards, Lizzette Lyla

内容来自 Freelancer News,版权归原作者所有。

阅读原文 ↗

把喜欢的内容,留在你的新标签页。

打开我的订阅