Vizion Kompjuterik Jobs

We need an AI-focused application that delivers 100% functionality in analyzing and solving geometric patterns in jewellery. The primary objective is to ensure precision and reliable output. The features include: 1. Detecting patterns across different jewellery parts such as any parts of boundaries, textures, cuts, or design elements (including broad patterns, sub-patterns, and intricate sub-to-sub patterns). 2. Identifying low-quality or unclear components within these patterns and seamlessly replacing them with high-quality elements from other sources. 3. Ensuring replacements naturally blend into the overall piece, maintaining structural integrity and visual coherence. The focus is entirely on achieving accurate functionality with results that meet expectations directly. Experienced f...

I’m building a computer-side accessibility layer that translates real-time hand movements, captured through an ordinary webcam, into system commands such as click, drag, scroll and shortcut activation. The goal is to give users with limited mobility an intuitive alternative to a mouse or keyboard, so recognition accuracy and low-latency response are essential. I already know the interface points I want to hit—virtual mouse events, configurable hot-keys and an on-screen overlay for feedback—but I need an experienced developer to design and train the hand-gesture recognition pipeline, wire it into the operating system (Windows first, cross-platform later), and leave me with clear documentation so the tool can be maintained and expanded. You’re free to pick the tech ...

Project Description We are looking for an experienced AI Video Analytics developer or company that already has an existing AI video analytics platform or can quickly develop a demo for client presentation. Important: We are not looking for a full development project from scratch at this stage. Our immediate requirement is a working demo or existing software that can be integrated with Matrix IP Cameras. If the demo is approved, the selected developer/team will be awarded the complete development and customization project. Immediate Requirement We prefer candidates who can provide either: An existing AI Video Analytics software with a live demo. A ready-to-use MVP (Minimum Viable Product). A prototype that can be completed within a short timeframe (7–15 days). A presentation-rea...

I am building a proof-of-concept that lets a Raspberry Pi follow a user’s gaze and immediately translate that data into visual output on an attached screen. At the end of the job I want to be able to look at a point on the display and watch the system draw a corresponding dot, line, or free-hand stroke in real time. What I already have • Raspberry Pi 4 with camera interface available • Freedom to choose the most suitable eye-tracking library (OpenCV, MediaPipe, PyGaze, or similar) • One HDMI monitor for testing, though the code should remain display-agnostic so I can later swap in a touchscreen or e-Ink panel without major refactoring. What I need from you • Set up and calibrate the camera-based eye tracker on the Pi • Translate raw gaze coordinat...

I’m building a computer-vision layer that can reliably recognise images and return structured results that the rest of my platform can consume. The core of the project is an AI-driven image-recognition model; I need someone who can take it from dataset preparation all the way to a callable endpoint. Here’s what success looks like for me: • A well-architected model (TensorFlow, PyTorch or comparable) trained on the dataset I’ll supply—or augmented with open-source sets you recommend—to reach an agreed accuracy threshold. • Clean, readable Python code with clear comments and a short README so future team members can retrain or fine-tune the model. • A lightweight API or Dockerised service that I can spin up on AWS/GCP and hit with a POST requ...

I have hours of raw fight, match, and court footage—boxing bouts, football games, basketball clashes, and MMA events—and I need those long recordings distilled into fast-paced highlight reels. Each reel should automatically surface the biggest moments: clean knock-outs, jaw-dropping goals, buzzer-beater threes, and decisive submissions, stitched together with sleek transitions and scoreboard overlays. I’m looking for a solution that leans on AI rather than manual scrubbing. Whether you prefer computer-vision toolkits (OpenCV, PyTorch, TensorFlow) or proven AI editors like Wisecut, Descript, or run-time FFmpeg pipelines, I’m open, as long as the end product is: • A 60–120-second, 1080p (or better) reel for each event I supply • Accurate, exciting ...

I want to launch a web-based platform that recognises sign language through a user’s webcam and instantly shows the corresponding letter, word, or short phrase on screen. The core aim is educational: helping deaf learners, their families, and teachers see whether a sign has been performed correctly and understand its meaning in real time. Here is what I have in mind. When someone visits the site, they grant camera access, perform a sign, and the page immediately displays the detected output along with a brief explanation of that sign. A clean, distraction-free interface, large typography, and high-contrast colours are essential so the information is easy to follow. Accuracy matters more than speed, but I still want recognition to feel smooth rather than laggy. I am open to the under...

I need a proof-of-concept mobile web application that works smoothly on both Android and iOS. The core idea is simple but technically rich: a visitor points their phone at a public monument, the camera captures an image, the app recognises the figures depicted, and then each of those historical characters speaks back in a distinct, period-appropriate voice. Here is the flow I have in mind: 1. Vision: Use a lightweight on-device or edge computer-vision model to identify every figure in the photo. Accuracy, speed, and the ability to handle outdoor lighting are crucial. No text recognition is required—just figure identification. 2. Voice design: For each identified figure, generate an era-authentic voice with ElevenLabs’ Voice Design API. Think marble statues suddenly speaking...

Build the computer vision system as if it were analyzing gameplay from a retail copy of the game. It must not rely on any internal game state, memory access, engine hooks, debug APIs, or hardcoded timing values. The system should make all decisions solely from the rendered video output, exactly as a player would see it on screen. The system must support two modes with a user-toggle: * Shot Meter Mode: Analyze the live shot meter visually in real time rather than using predefined release timings or static values. * Animation Mode: Analyze the player’s shooting animation to determine the optimal release timing without using the shot meter. The solution should process frames with low latency, be resilient to different camera angles, graphics settings, and jump shots, and include a ca...

• Description We will provide the sports data to you. Screen video frames will be fed into a backend engine via API: 1. Types of sports What sport is it? Football/basketball/rugby/American football, etc 2. List of Logos Detected logos: UEFA, Real Madrid, Bayern Munich, Liverpool, Inter, etc 3. List of players Identify players from the shirt: 10 Kylian Mbappe, 11 Rodrygo, 12, 13 … 4. Home/away From the players’ wears: Real Madrid home/away, Liverpool home/away etc 5. Match info (team and score) on the top left corner of the screen (if there is any) 6. Who is holding the ball. Upon a worker's initiation and processing, your engine will store the identified information in memory and send it back to the front end. The worker can be re-initiated/reused. For train...

Development of an OCR-Based Industrial Vision System using NVIDIA Jetson Orin Nano About Us We are an industrial automation company specializing in robotic palletizers, case packers, end-of-line automation and custom machinery. We are building a reusable machine vision platform, and this OCR project is Phase 1. Project Objective Develop a robust OCR application for NVIDIA Jetson Orin Nano using a camera to read printed alphanumeric characters. The architecture should be modular and reusable for future barcode, QR, AI inspection and robot guidance applications. Hardware Platform  NVIDIA Jetson Orin Nano Developer Kit  Raspberry Pi HQ Camera (Sony IMX477) or equivalent CSI camera  LED illumination  C-Mount Lens  Future migration to Basler industrial camera Scope of Work Camera Integrat...

We are looking for freelancers to contribute to a large and diverse facial image dataset that will support machine learning research and facial recognition model training. The project focuses on collecting high-quality facial images that represent real-world variations in identity, head poses, appearance, and age progression. Each freelancer will complete four image collection tasks, including front-facing neutral captures, controlled head-pose images, and historical facial images showing natural changes over time. Freelancers are required to submit up to 28 approved images, with a minimum of 11 valid images needed for an approved and billable submission. Images must be authentic, high quality, and reflect natural variations in clothing, hairstyle, lighting, background, facial angles, a...