Viziune computerizată Jobs

I want a single, turnkey application that watches my CCTV feeds, spots shop-lifters in real time, recognises grocery products on the shelves, and keeps a live head-count of customers. The core model must be YOLO, and I need the exact same code-base to compile and run on both Windows (desktop with NVIDIA GPU) and a Raspberry Pi 4. Video sources vary—some cameras stream RTSP over IP while a few older analog units reach the NVR through a capture card—so the program has to accept either type without manual re-configuration. For product recognition I care only about groceries; no clothing or electronics labelling is necessary. The model should be trained (or fine-tuned) on the most common supermarket items so false positives stay low even when shelves are crowded. Key expecta...

I'm looking to develop specialized software for an artillery correction system using a PTZ (pan-tilt-zoom) camera. The software should be capable of detecting and tracking static targets. Additionally, it should provide crucial correction data such as angular position adjustments and elevation adjustments. Key requirements: - Target detection and tracking of static targets - Provide angular position adjustments - Provide elevation adjustments Ideal skills and experience: - Expertise in computer vision and image processing - Experience with PTZ camera integration - Strong background in software development, particularly in real-time systems - Familiarity with artillery systems and requirements is a plus

I need an end-to-end deep-learning model that can pick out identical human faces across images and video in real time. The core requirements are straightforward: • Detect every human face in an image or live stream, draw accurate bounding boxes, then compare each face against a gallery to decide whether it is an identical match. • Serve three environments without extra rewrites: security-grade CCTV feeds, social-media style mobile uploads, and large photo-management archives. • Deliver low latency on a single modern GPU while still running acceptably on CPU-only hardware for lightweight deployments. I’m comfortable with either PyTorch or TensorFlow/Keras; use the framework you know best. A pre-trained backbone such as ResNet, MobileNet, or Vision Transformer is...

Title: Senior Android Developer: Hybrid Computer Vision Pipeline (Local ML + Gemini API) Project Description: We are seeking a Senior Android Engineer to develop a modular, high-performance infrastructure for real-time video capture and analysis. The core of the project is a "Hybrid AI Routing Engine" that intelligently switches between on-device local processing and high-depth cloud analysis. This application is an R&D prototype that must be Play Store Ready, with a heavy focus on background stability, thermal management, and seamless connectivity transitions. Technical Specifications: 1. Camera Pipeline & Intelligent Routing: Implement CameraX to ensure maximum compatibility across various Android OEMs (Samsung, Pixel, Xiaomi, etc.). Develop a Routing Engine to ing...

I'm looking for someone to develop an App similar to Cal Ai, SnapCalorie and Qalzy. I work in the fitness industry and would like to sell the app to online customers. If you think you can develop something similar, please reach out and we can discuss the details.

I need a Python-based tool that can scan video files, detect Trees, Road, Infrastructure—and leave room for many more object classes as the project grows—then save every relevant frame straight into a MySQL database. Each stored image must be linked to the corresponding frame number and any detection metadata so I can later query, filter, and analyse the results. Once the data is stored, I want a lightweight viewer that steps through the saved frames in order, overlaying the detection boxes so I can visually confirm accuracy. OpenCV for frame extraction and display is acceptable; YOLO, TensorFlow, or another modern model is fine so long as the code is clean, well-commented, and easy for me to retrain with additional classes. To keep the hand-off smooth, please include: •...

I need a researcher who can build a production-ready model that listens to a baby’s cry, watches the paired video, and decides—reliably—whether the cause is hunger, discomfort, or simple attention seeking. Audio and video must be fused inside one architecture; running them in parallel but independently will not satisfy our accuracy goals. You may use the deep-learning stack you trust most (PyTorch, TensorFlow, Keras, OpenCV, torchaudio, etc.) provided the final network can run in real time on an edge device and be exported to ONNX or TFLite. I will share product constraints and a small proprietary data set; you will expand it through public sources or augmentation, perform rigorous cross-validation, and refine the model until we consistently exceed 90 % precision and rec...

I need an image classification AI model to run on a Raspberry Pi 4. This model will detect whether individuals are wearing PPE kits in static images. The primary application will be in industrial sites. Key Requirements: - The model should accurately classify images as 'PPE' or 'No PPE'. - It needs to be optimized for performance on Raspberry Pi 4. - The input will be static images, not live feeds or videos. Ideal Skills and Experience: - Proficiency in AI and image classification techniques. - Experience with model optimization for edge devices, especially Raspberry Pi. - Background in developing AI solutions for industrial applications is a plus. - Ability to deliver a robust and efficient model within budget constraints.

Want to add custom commands upload to robot and from retroit control execute that commands

Autonomous Agricultural Robot Developer (ROS2, Jetson Nano) – Short Freelancer Posting We are developing CropScanalyzer, an autonomous agricultural rover for crop phenotyping and plant health monitoring. We are looking for an experienced ROS2 robotics developer/team for autonomous navigation, multi-motor control, and sensor/data integration. Robot Details: • 4-wheel drive + 4-wheel steering (8 motors) • Additional motors for conveyor plant row deflector + camera vertical motion (~10 motors total) • NVIDIA Jetson Nano onboard computer • Sensors: RTK-GNSS, LiDAR, IMU-depth cameras, GoPro cameras, encoders, light, and load sensors. Key Tasks: • ROS2 autonomous navigation with sensor fusion • Manual remote-control mode + fully autonomous path-following mode ...

I need a small application that, once given a set of latitude-longitude boundaries for a desert sector, we can use older maps for any available dates with high resolution. stores it locally, and then runs an object-detection model that flags cars and trucks only. The moment a vehicle silhouette is spotted, the program must return its exact coordinates so rescuers can be dispatched quickly. I’m flexible about the imagery source—NASA, ESA, Google Earth, or any other free feed is fine as long as it delivers cloud-free, high-resolution scenes. You can use the tool to capture screenshots by moving in circles around the selected location. The detector has to work at desert scale, so please build it with an established computer-vision framework (e.g., TensorFlow, PyTorch, YOLO, or ...

I’m building a video-surveillance module and need a Convolutional Neural Network that can spot humans, vehicles, and animals the instant they appear on-screen, whether the cameras are indoors, outdoors, or a mix of both. As soon as the model flags one of those classes, it must immediately push an alert to my back-end (REST webhook is fine) and simultaneously initiate recording on the camera stream. Speed is critical: I’m targeting sub-100 ms inference per frame on an Nvidia Jetson Xavier, yet I still need accuracy good enough to avoid nuisance alerts in busy scenes. You’re free to choose the framework you prefer—YOLOv8, Faster R-CNN, or a custom TensorFlow / PyTorch implementation—as long as the final package runs headless in Linux and can be containerised (D...

I have a curated dataset of abdominal X-ray images that needs a robust deep-learning model capable of classifying key clinical findings. The end goal is a production-ready Python solution that can consistently score above 90 % accuracy on an unseen validation set. You’ll start with any mainstream framework you prefer—TensorFlow, Keras, or PyTorch—and handle the full pipeline: data preparation and augmentation, model architecture selection, training, hyper-parameter tuning, and evaluation. Please keep the code modular and well-commented so I can retrain or fine-tune later as new data comes in. A concise report that explains your decisions, metrics, and suggestions for future improvements will also be appreciated. To help me choose quickly, focus your proposal on your exp...

Computer Vision Prototype Needed — Head Tracking & Auto Zoom (R&D Phase) I’m building a product and am currently in the early R&D phase before final hardware and SDK delivery. We are looking for an experienced computer vision / mobile vision developer to help us prototype and validate: Core Features • Real-time head/face detection • Automatic head tracking (keeping the head centered) • Auto zoom / smart cropping (keeping head size consistent) • Smooth motion and transition logic At this stage, we do not yet have final hardware or SDK access, so we are asking the freelancer to: Phase 1 – Prototype (on your own hardware) • Build a working demo using your own Android device / webcam / test camera • Use open-source tools (MediaPi...

I'm seeking an experienced AI developer to create a computer vision model focused on detecting people. The model will need to function effectively in both indoor and outdoor environments. Key Requirements: - Primary function: Object detection with a focus on people - Adaptable to both indoor and outdoor settings - High accuracy and reliability Ideal Skills and Experience: - Expertise in AI and machine learning - Strong background in computer vision, particularly in object detection - Experience with datasets and training models for varied environments - Proficiency in programming languages such as Python, and familiarity with libraries like TensorFlow or PyTorch Please provide examples of similar projects you've completed.

[FFmpeg-Based Video Processing Specialist] We are looking for a specialist who can handle video processing within an automation pipeline and configure basic subtitle and animation settings. Main Responsibilities: - Install and configure FFmpeg - Set up automated video conversion and clip editing - Configure basic subtitle style (font, position, timing) - Apply simple animation effects - Analyze error logs and improve system stability (Detailed tools and internal structure will be shared after meeting)

We have a platform that requires user to do a face capture, means user will take a live picture of themselves The goal is to ensure that only real users use our system. Let me know if you have done anything like this or capable

I want to build a Windows-only application that watches the live screen, detects people—including stylised in-game characters—in real time, and immediately drives a game controller so the crosshair locks onto and tracks that target as it moves. The workflow I picture is straightforward: the program captures frames, an AI model spots the person, and a virtual stick signal (XInput, vJoy or a comparable driver) nudges the aim every frame so it stays centred. Smoothness and speed are critical. On a 1080p feed I’m aiming for roughly 60 fps with no more than 40–50 ms end-to-end latency, so techniques such as YOLOv8, TensorRT or a lightweight custom network combined with OpenCV screen capture should fit. You’re free to choose Python, C++, or another language as long...

More details: What type of fashion tech MVP are you looking to build? virtual try on with fit prediction What key features do you want in your virtual try-on system? Size and fit prediction What type of size and fit prediction functionality do you need? Measurement-based predictions, User profile customization

We are building an AI-powered livestream production platform that can fully automate the broadcasting of sports competitions, starting with dog agility. The goal is to eliminate the need for camera operators and directors, while still delivering a professional, TV-like livestream using one or multiple cameras. This is not a theoretical idea – similar systems exist (Pixellot, PlaySight), but none are specialized for agility sports nor integrated with timing systems and live graphics. Our project focuses on rule- based AI direction + computer vision, not generative AI. MVP Scope Video & Livestream: - Capture video from 1–2 cameras (USB / IP / phone stream) - Output live RTMP stream (YouTube / test server) - Automatically switch cameras based on AI decisions AI Director Logi...

I have a small batch of images—fewer than one-hundred—that need clean, consistent object-detection labelling. For each image you will draw tight, non-overlapping bounding boxes around every instance of the target classes I will supply once we start. Accuracy matters more than speed; missed objects or sloppy boxes will be rejected. Preferred workflow is any modern tool that can export to COCO JSON or Pascal-VOC XML, as these formats plug straight into my training pipeline. If you normally use LabelImg, CVAT, Supervisely, or similar, that’s perfect. Deliverables • Annotated dataset in COCO JSON or Pascal-VOC XML (your choice, just stay consistent). • A quick text report summarising class counts and any edge cases flagged during labelling. I will run a...

I have a set of roughly 51 – 200 technical diagrams that must be fully prepared for computer-vision training. Each image needs three things: clear class labels, accurate bounding boxes around every object of interest, and pixel-level segmentation masks. You may work in any modern annotation platform you prefer—CVAT, Labelbox, VGG Image Annotator, or a comparable tool—as long as the final export is delivered in COCO JSON (or another widely used format we agree on before you begin). Consistency is critical; every diagram must follow the same labeling taxonomy and colour scheme so the dataset can be dropped straight into a model pipeline without extra cleaning. Deliverables • Labeled, boxed, and segmented versions of each diagram (51–200 images) • One...