Browse categories
Explore
Fiverr Pro
English
$
USD
I will train a custom yolo object detection model on your dataset
Maria
About this gig
Need a model that finds YOUR objects not the 80 generic COCO classes?
I train custom YOLO detection models on your images and hand you something you can actually run.
WHAT YOU GET
Trained model weights, ready to use
Clean inference script run on an image, folder, video or live webcam
Accuracy report: mAP50, mAP50-95, precision, recall, confusion matrix
Screen recording showing exactly how to run it
Full source code, full commercial rights
WORKS FOR
Defect detection PPE and safety compliance Retail shelf counting Vehicles and licence plates Medical imaging Crop disease Drone and satellite imagery Anything you can photograph
NO LABELED DATA? I annotate for you. Just ask before ordering.
BACKGROUND
5+ years in AI. MS in Computer Science. 3 peer-reviewed papers, including a medical imaging study reaching 97.4% accuracy on MRI data.
I'll tell you up front if your dataset is too small, instead of taking the order and delivering a 40% accuracy model.
MESSAGE ME FIRST with what you want to detect and roughly how many images you have. I'll give you an honest accuracy estimate before you order.
Get to know Maria
Maria
AI Engineer,RAG Chatbots, AI Agents, Computer Vision, ROS Robotics
- FromPakistan
- Member sinceJun 2022
- Avg. response time1 hour
- Last delivery2 years
Languages
Urdu, English, Hindi
I build AI systems that see, understand and act.RAG chatbots and AI agents that answer from YOUR documents, computer vision models that detect what matters in your images and video, and ROS2 robots that navigate on their own.Delivered across healthcare, agriculture, legal and e-commerce.Stack: LangChain, LlamaIndex, OpenAI + Claude API, Ollama (private, on-premise), Pinecone, ChromaDB, FastAPI — plus YOLO, OpenCV, PyTorch, ROS2 and Gazebo.Tell me what you want to build and I'll say honestly whether it's realistic.
My Portfolio
Other AI Development Services I Offer
FAQ
How many images do I need?
Minimum 100 per class for something usable, 300+ per class for production accuracy. Send what you have and I'll give you an honest estimate before you order.
My images aren't labeled. Can you help?
Yes. Annotation is included in Premium, or available as an extra on the other packages.
What accuracy will I get?
It depends almost entirely on data quality. Clean, varied images of visually distinct objects typically reach 85-95% mAP50. I give you my estimate before you order, not after.
Can it run in real time on a Jetson, Raspberry Pi or phone?
Yes. Tell me your target device before ordering so I choose the right model size and export format.
Can it run in real time on a Jetson, Raspberry Pi or phone?
A: Yes. Tell me your target device before ordering so I choose the right model size and export format.
Detection or classification, which do I need?
A: Detection finds where objects are and draws boxes. Classification says what an image is as a whole. Describe your problem and I'll tell you which fits.
Do I own the code?
100%. Full source, full commercial rights, no licence fees, no dependency on me afterwards.
