Browse categories
Explore
Fiverr Pro
English
$
USD
I will train a custom yolo object detection model on your dataset
About this Gig
Need a custom object detection model that works on your own images? I will train a YOLO model on your dataset and deliver a model you can use right away.
I'm a Computer Engineer specializing in computer vision. My recent project: a Turkish traffic sign detection system for an autonomous vehicle team competing at Teknofest. I benchmarked 9 YOLO models (v8, v11, v12) and reached 0.91 mAP@0.5 on 29 classes.
What you get:
- Dataset check (label errors, class imbalance, duplicates)
- YOLO training with Ultralytics (YOLOv8 / v11 / v12)
- Trained weights (.pt) ready for inference
- Clear report: mAP, precision, recall, confusion matrix
- Inference script to run the model on images or video
- ONNX export and model comparison (Premium)
Use cases: product detection, quality inspection, traffic and vehicle detection, safety equipment, agriculture, retail, and more.
What I need from you:
- Labeled images (YOLO, COCO or Pascal VOC format), or unlabeled images plus the labeling extra
- A short description of your use case
Please message me before ordering so we can confirm your data and pick the right package.
Expertise:
Image processing
•
Classification
•
Object detection
•
Other
Programming language:
Python
•
Colab
Tools:
OpenCV
•
Colab
•
PyTorch
Frameworks:
PyTorch
•
Other
My Portfolio
FAQ
My images are not labeled. Can you still help?
Yes. You can add the "Data labeling" extra, and I will annotate your images with bounding boxes in YOLO format. For large or complex datasets, message me for a custom offer.
How many images do I need?
As a rough guide, 100–300 images per class give a solid starting point. Fewer can work for simple objects. I will check your data and tell you honestly if more is needed.
Can you guarantee a specific accuracy?
Accuracy depends mostly on data quality and quantity, so I can't guarantee a fixed number. I will always explain the results, point out weak classes, and suggest how to improve them.
Which YOLO version will you use?
I usually compare YOLOv8, v11 and v12 and pick the best trade-off between speed and accuracy for your use case. If you need a specific version, just let me know.
Can the model run on edge devices or in real time?
The Premium package includes ONNX export, which works on most platforms. For specific hardware (Jetson, Raspberry Pi, etc.), message me first so we can discuss your requirements.
What is the demo app in the Premium package?
A simple local web interface (built with Gradio) where you can upload images or videos and see the model's detections. It runs on your own computer with one command. Hosting it online or building a full web/mobile app is not included.
