Shelf Item Counting POC
$250-750 USD
Posted 1 minute ago
•
Ends in 6 days
$250-750 USD
Paid on delivery
I need a compact proof-of-concept that can look at a single photo of a store shelf and return one simple number: the total quantity of products it finds. No bounding boxes, no section breakdowns—just the grand total per image displayed on a lightweight web dashboard. You will receive a small batch of sample shelf photos to start with. I’m open to whichever computer-vision stack you prefer (Python, OpenCV, TensorFlow, PyTorch, YOLO, etc.) as long as the final flow is clear and reproducible: • I upload or drag-and-drop an image. • Your model processes it and sends back the count. • The dashboard records and displays that number next to the image thumbnail. Accuracy does not need to be retail-grade yet, but the proof must show that the pipeline works end-to-end and can be fine-tuned later for brand recognition or section-level analytics if we extend the project. Deliverables: 1. Source code with brief setup notes (virtualenv or Docker preferred). 2. Running demo (local or small cloud instance). 3. Short read-me summarising model choice, training approach, and areas for improvement. I’ll test using a separate set of photos; if the dashboard reports a reasonable count within a few seconds per image, the POC is accepted.
Project ID: 40814545
About the project
22 proposals
Open for bidding
Remote project
Active 6 secs ago
Place your bid
Bid amount Email addressBenefits of bidding on Freelancer
Set your budget and timeframe
It's free to sign up and bid on jobs
22 freelancers are bidding on average $390 USD for this job


Hello, I can help you with "Shelf Item Counting POC" as per your given project description. We can discuss more in detail during a chat conversation when you are available. I've worked on many PHP projects in recent times. So I am confident on achieving your expected Goals. Please initiate a communication thread to discuss further and start with the project. ⭐ 5.0/5 from a recent client: "Excellent work. Would definitely utilize again. Stellar job!" Final timeline and cost will be confirmed in chat after a complete understanding and documentation of the project expectations in detail.


Hi there, You need a proof-of-concept that counts products in a single shelf photo, displaying the result on a lightweight web dashboard. I have experience building similar models using Python and OpenCV, which makes this a familiar task for me. How I would approach it: - Set up a streamlined pipeline where images can be uploaded via drag-and-drop. - Develop a computer vision model using OpenCV or TensorFlow for counting products in the photo. - Integrate the model with a simple web dashboard that displays the count next to the thumbnail. - Provide a running demo locally or in a small cloud instance, with source code and documentation included. I have over four years of experience in computer vision projects, effectively documenting setup and processing for easy handover. Each step will be clear, allowing for future improvements as needed. What types of images will you be providing for the initial batch? This will help me align the model training accordingly. I can deliver the proof-of-concept in two days, showing a clear and functional pipeline.


Hi there, Laravel and PHP are my main stack, including PHP and JavaScript. I have built SaaS dashboards, admin panels and payment workflows in Laravel. For "Shelf Item Counting POC" I would write clean, secure code that is easy to maintain and extend. How I would approach it: - Review the current code, or plan the structure from scratch - Build the features in clear milestones with regular updates - Test, deploy and hand over documented code Is this a new build, or are we working on an existing Laravel project? Message me and I will share a similar Laravel project I delivered. Umair


IF YOU ARE NOT HAPPY YOU DON'T PAY! A stock COCO model barely has a "product" class. On a packed shelf, its NMS step also merges neighbouring identical packs, so counts come out low. I'd start from a YOLO detector trained on SKU-110K, a dataset of dense retail shelves, and check it against your sample photos before fine-tuning anything. You'd get a Dockerized Python service with a drag-and-drop dashboard that shows each thumbnail next to its count, plus the readme on model choice and next steps. Should the number count only the visible front-facing items, or do you expect an estimate of stock hidden behind them? Send me a message and I'll explain exactly how I'd approach your project.


Counting items from a single shelf photo is really an object detection problem. I would fine tune a YOLOv8 model on a small labeled set and return the item count plus bounding boxes so you can verify accuracy visually. Working POC in 4 days using Python and OpenCV for preprocessing. These numbers are based on the post as written, we will refine both after a quick scope talk. Can start today. Want me to send a quick plan?


Hi, I’ve reviewed your Shelf Item Counting POC and understand what you’re looking to achieve. I can help you handle the required work with a clean, reliable, and professional approach. I have experience with PHP, JavaScript, Python, Django, OpenCV, Tensorflow, Pytorch, Computer Vision and can work effectively with your existing setup, requirements, and workflow. I’ll focus on proper implementation, responsive functionality, attention to detail, and thorough testing to make sure everything works as expected. I’ll also make sure the completed work is aligned with your requirements rather than just providing a quick or temporary solution. I’m available to start immediately and can work according to your required timeline. Best regards, Khadija Tul Kubra


A single image shelf item counting proof of concept that uploads a photo, runs a vision model and returns the total product count on a lightweight web dashboard. I will build a FastAPI backend that receives the image, processes it with a pretrained YOLO model fine‑tuned on your sample set, and outputs a single integer. The frontend will be a minimal React page with drag‑and‑drop upload, thumbnail preview and the reported count displayed beside it. I will containerize the whole stack with Docker, include a PostgreSQL table to log each image and its count, and provide a concise README with setup steps and how to replace the model for later brand or section analysis. The source will be clean, documented and ready to run locally or on a small cloud instance. Do you prefer the model training data to be stored locally or accessed from a cloud bucket? Let’s chat, lock in the exact scope and get started.


Hello, "Store Shelf Product Counting POC" - the goal is a working upload-to-count pipeline, not an overcomplicated analytics platform. I'll build the lightweight dashboard in JavaScript, connect it to a Python processing backend, and use Docker for a reproducible setup. Each upload will be processed, assigned a total count, and saved alongside its thumbnail. I'll also document the model choice and test results. One important limitation to address early is overlapping or partially hidden products, which can cause undercounting even when the image pipeline works correctly. Would you be open to using an existing pretrained detection model rather than training a custom model from scratch for this first POC? Regards, Beate


Hello, You need a simple shelf-counting tool that takes an image and spits out a product total, no frills, just the number on a dashboard. The tricky part here is balancing speed and accuracy with minimal training data, since you're starting with a small batch of sample images and need something that gives reasonable counts fast. I'd start by testing a few lightweight object detection models (likely YOLO or a custom OpenCV approach) to see which handles your shelf images best without overcomplicating it. Once we have decent counts, I'll wire it into a basic Flask or FastAPI backend with a drag-and-drop UI that logs the results. The whole thing can run in Docker for easy setup. Did something similar last month, a document scanner that extracts field counts from forms using AWS Textract, so the pipeline setup is fresh. How strict is "a few seconds per image", are we talking under 3, or is 5-6 acceptable while we tune it?


Extracting a single item count from a shelf photo without getting bogged down in bounding boxes makes this a clean proof of concept to build. I would set up a Python backend using OpenCV or a lightweight YOLO approach to process the uploaded image, extract the total count, and pass it directly to a minimal web dashboard. You get a working drag and drop upload flow that returns the grand total next to the thumbnail within seconds. Let me know if you have a sample batch of photos ready to test with.


Hello, "Single-Image Shelf Product Counting" - turn each uploaded shelf photo into one product count, recorded beside its thumbnail on a simple dashboard. I'll build the upload and processing flow in Python, use an image-analysis pipeline to estimate the total product count, and connect the result to a lightweight web dashboard with image history. I'll package the source and setup notes in Docker, document the model's limitations and fine-tuning path, and personally handle implementation and testing end to end. Can you provide a few sample shelf photos and manually verified product counts to use as the initial test set? Looking forward to working with you. Aleksandar


Hello, I have gone through "Shelf Item Counting POC" and the PHP requirements are clear to me. How I would approach it: - Review your existing code or requirements and agree the scope and milestones with you. - Build or fix the features in clean, secure PHP with a well structured MySQL database, keeping your current functionality intact. - Test thoroughly, deploy, and hand over with brief documentation of the changes. I have delivered many PHP projects, including custom systems, admin panels and fixes to existing code. Could you share the current code or access, and any details of what should change? Tarim


Hello, I read your Shelf Item Counting POC brief and I understand you want a compact proof of concept where a single shelf photo is uploaded and the system returns one simple number the total product count displayed on a lightweight web dashboard next to the image thumbnail with no bounding boxes or section breakdowns. My approach is to build this in Python using YOLO or a comparable open source detector with OpenCV for image handling and a simple Streamlit or FastAPI dashboard focused on a clear reproducible flow. The model will process each image count the visible products and pass the number to the dashboard which records and displays it alongside the thumbnail. I will keep the pipeline light so each image returns a count within a few seconds and structure the code so it can later be fine tuned for brand recognition or section level analytics without rewriting the core. You will receive the source code with brief setup notes using a virtualenv or Docker a running demo local or on a small cloud instance and a short read me covering model choice training approach and areas for improvement. I would love to review your sample shelf photos so I can choose the right model and confirm the expected accuracy level for the POC. Best regards, Yasir Ahmed.
内容来自 Freelancer News,版权归原作者所有。
阅读原文 ↗