Computer vision service

Dear client,
I have solid experience in computer vision and image-based object detection, especially for identifying structural elements such as windows and sliding doors. For your project, I will build a reliable detection model, analyze each object, calculate the number of sections, and estimate the object area with high accuracy. I will deliver clean code, detailed documentation, and stable results.
Thank you for your consideration, and I look forward to working together.

Hello, the project is quite interesting.
Technically, detection and counting of sections can be implemented. The main question is in calculating the area; from the photo, it is impossible to understand the scale. For example, a window sized 2x2 meters, taken from 10 meters away, occupies the same number of pixels as a window sized 1x1 meter, taken from 5 meters away. Therefore, it is not advisable to rely on this indicator.
For training the first version of the model, I will need as many photos as possible from you of different types of windows and doors that will be used in the service.
I want to know more about how you envision the logic of using your future service and what kind of photos will generally be submitted to it (poor average quality photos or more professional interior shots).

Show original

Translate

Good day!
I am Valentin from Arctic Web. We are engaged in IT consulting and full-cycle development: from UX/UI design to web and mobile development and integrations. We are ready to take your project into work and share examples of similar cases in personal messages.

Sincerely,
Arctic Web Team
Freelancehunt

Show original

Translate

I am ready to complete this project and have extensive experience in developing various applications.

Show original

Translate

Hello Anton,

I have analyzed your request for a window/door detection service. While object detection (identifying a "window") is standard, the requirement to calculate Area and count Sections presents a specific computer vision challenge that requires more than just a basic AI model.

As a PCAP™ Certified Python Developer and CS student with a focus on Full Stack and AI, here is the technical approach I propose to ensure accuracy:

1. Object Detection (Windows & Doors)
I will utilize YOLOv8 (You Only Look Once) trained on a custom dataset of architectural elements. This is superior to generic APIs (like Google Vision) because we can fine-tune it specifically to distinguish between "Sliding Doors" and "Fixed Windows."

… 2. The "Area" Challenge (The Physics Problem)
A 2D image does not have inherent scale. To calculate the area ($m^2$) accurately, we cannot just count pixels. I will implement two options for you:

Reference Mode: The user places a standard object (like a coin or A4 paper) or defines a reference height (e.g., "Assume door is $2.1 \text{ m}$").
Homography/Perspective Correction: Using OpenCV to correct the angle if the photo is taken from the side, ensuring the calculated area doesn't skew.
3. Counting Sections (The detailed analysis)
Once the window is detected, I will run a secondary OpenCV Edge Detection algorithm (Canny/Hough Lines) within the bounding box to count the vertical/horizontal dividers (mullions). This allows us to accurately report: "1 Window, 3 Sections."

The Deliverable:
A Full-Stack Web Service (Next.js Frontend + Python FastAPI Backend) where you upload an image and get a visual overlay + a JSON report of the data.

I am ready to build a Proof of Concept (PoC) for the detection logic immediately.

Best regards,

Revaz Goguadze

Good day, I will develop a service for detecting necessary objects in images with data analysis of these objects, feel free to contact me, I will be happy to collaborate.

Show original

Translate

Привіт, Антон.
У мене вже є виконане завдання. Я його втілював для іншої компанії. Там це все втілено на платформі N8N. Інтеграція сервісу Mistral, яка виступає в ролі розпізнавання об'єктів на зображенні їхнім детальним описом. Також це можна втілити через Google Vision та через OpenAI. Тут вже в залежності від деталізації, яка вам потрібна, і подальше передачі. Це все можна реалізувати в короткі терміни за допомогою моєї логіки. Тому звертайтеся. Результати можете переглянути в моєму портфоліо. Його можна адаптувати під різні типи. Дякую. Звертайтеся, буду чекати на повідомлення.

Hello, Anton.
I already have a completed task. I implemented it for another company. It is all implemented there on the N8 platform. Integration of the Mistral service, which acts as the recognition of objects in the image with their detailed description. This can also be implemented through Google Vision and through OpenAI. It already depends on the level of detail you need, and subsequent transfer. All this can be implemented in a short time using my logic. So, get in touch. You can view the results in my portfolio. It can be adapted for different types. Thank you. Get in touch, I will wait for a message.

Hi there!

I can build the image tool so a user uploads a photo and sees windows or sliding doors marked right on the image. Below the picture, there will be clear results showing the number of objects, their sections, and approximate areas.

To keep area math simple but more accurate, I'll add a small built-in helper that reads one reference point from the photo, like a hand or wall corner, so the service scales its estimates without needing measurements. That keeps the flow fast and human-friendly.

Thank you!

Hello, Anton

I can make this code on python.
Budget and deadlines discussed in PM.

Hello,
I'm a Python developer.
I've completed many similar projects.
I'm ready to take on new work and complete it quickly and efficiently.
I deliver high-quality work on time.
You can read the reviews.

Hello, I can develop a service in Python that accepts photos, identifies windows and sliding doors in them, and also counts the number of sections and the approximate area of the objects.

For detection, I use: Python (YOLOv8/YOLOv9, OpenCV, PyTorch.)
For the backend: Python (FastAPI or Flask.)

I can create this as a PWA application that will work as a mobile app directly from the browser and will allow users to take photos and send them for analysis.

Show original

Translate