Door categorieën bladeren
Ontdekken
Fiverr Pro
Nederlands
$
USD
I will build yolo object detection and tracking for images and video
AI Software Architect, AI SaaS, RAG, Agents and Automation
I'm Hussain, founder of Afaxon AI and Software Architect. I build end-to-end AI SaaS products, RAG systems, AI agents and custom AI automation.
I understand your system, clarify requirements and desi...
Niveau 2
Voldoet aan hoge prestatiecriteria en heeft een bewezen staat van dienst in het voldoen aan de verwachtingen van de klant.
Over deze dienst
Detect objects in images and follow them through video with a custom YOLO solution. I build object detection and tracking pipelines for products, vehicles and other agreed object classes.
Using Python, PyTorch and OpenCV, I prepare your labeled data, fine-tune a detector, evaluate it on held-out examples and deliver runnable code. Tracking adds object IDs across frames for one video source.
BASIC: YOLO detector; up to 500 labeled images and 3 classes.
STANDARD: Detection plus tracking; up to 2000 labeled images, 5 classes and 1 video source.
PREMIUM: Detection and tracking; up to 5000 labeled images, 10 classes, 1 source, 1 API and deployment to 1 agreed environment.
All packages include model weights, source code, evaluation results and setup documentation. Larger datasets, multiple cameras and C++ integration need a custom scope.
Send sample images/video, class names, annotation format and target hardware before ordering. Accuracy and speed depend on data and hardware; targets are agreed after review. Annotation, compute/hosting fees and ongoing monitoring are separate. Segmentation, OCR and general image enhancement are outside this Gig.
Programmeertaal:
Python
•
Colab
•
Overige
Tools:
Jupyter-notitieboek
•
opencv
•
Colab
•
PyTorch
Frameworks:
PyTorch
Mijn portfolio
Veelgestelde vragen
Do I need an annotated dataset, and what are the limits?
Yes. Supply labeled images: Basic up to 500 images/3 classes; Standard 2000/5; Premium 5000/10. Send samples first so I can assess quality, format and suitability. Annotation, data collection and larger datasets require separate scope.
What files and deployment work are included?
All packages include model weights, source code, evaluation results and setup documentation. Standard adds tracking for one video source. Premium adds one API and deployment to one agreed, ready environment. Compute and hosting fees are separate.
What is the difference between detection and tracking?
Detection identifies objects and their locations in each image or video frame. Tracking links detections across frames using object IDs. Tracking is included in Standard and Premium for one agreed video source. It does not identify a person by name.
Can you guarantee a specific accuracy or real-time speed?
I evaluate performance on held-out examples and report results. Accuracy, tracking stability and speed depend on data, lighting, occlusion, resolution and hardware. Send your targets and sample data first; I do not promise an untested accuracy or frame rate.
Do you include segmentation, OCR or other OpenCV services?
This Gig focuses on object detection and tracking. Pixel-level segmentation, text recognition/OCR, image enhancement and unrelated OpenCV tasks require a separate service or custom scope.
Can you handle complex camera or C++ integrations?
Yes, I can assess custom camera/video pipelines and Python or C++ integration. The listed packages cover one agreed source; multiple cameras, specialized hardware, C++ integration and larger systems need a custom offer after technical review.
What does one video source and deployment environment mean?
One source is one agreed video input, such as a file-processing pipeline or a supported camera stream. Share sample clips, resolution, format and connection details before ordering. Premium covers one ready environment; hardware procurement and ongoing hosting are excluded.
How do training, testing and revisions work?
We agree the model setup, compute budget, dataset split and acceptance examples before ordering. Revisions refine the agreed classes and pipeline: 1/2/3 rounds for Basic/Standard/Premium. New classes, data collection or added integrations require separate scope.
