I will develop yolov8 object detection and segmentation models
About this gig
I'm Bhavya, a Data Science grad student in NYC specializing in computer vision pipelines using YOLOv8, PyTorch, and OpenCV. I build systems that work in production not just demo scripts.
My last YOLOv8 project: mAP50: 0.995 | Precision: 0.998 | Recall: 1.00
What I can detect/segment:
- Products, defects, or anomalies (manufacturing/QA)
- People, vehicles, or objects (surveillance/safety)
- Medical or scientific imagery
- Custom classes if you can label it, I can train it
What you'll get:
- Trained model weights (.pt file)
- Full validation report (mAP, Precision, Recall, Confusion Matrix)
- Inference script (Python) run predictions on new images instantly
- Clean documentation so you can actually use what I deliver
Standard & Premium also include:
- Manual annotation via LabelMe/Roboflow
- Streamlit app for real-time inference
- Deployment guidance
Before ordering: Message me with your dataset size, number of classes, and what you're trying to detect. I'll tell you honestly which package fits.
Let's build something that actually works.
Get to know BhavyaP
Python and Data Specialist, Let's get to work
- FromUnited States
- Member sinceDec 2025
- Avg. response time4 hours
- Last delivery3 months
Languages
English
FAQ
How do you handle the dataset?
You provide raw images - I handle manual labeling via LabelMe or Roboflow setup depending on your package. Training runs on my local GPU or your cloud instance for large-scale datasets.
Do you offer Instance Segmentation?
Yes. I deliver pixel-perfect polygon masks using YOLOv8/v11 - far more accurate than bounding boxes, and essential for medical, industrial, or retail use cases where precision matters.
Is the UI included?
A custom Streamlit dashboard is included in the Premium package. Basic and Standard packages include trained weights (.pt/.onnx) and ready-to-run inference scripts.
Can the model run in real-time on my hardware?
Yes. I optimize for your specific setup - NVIDIA GPU, Jetson, or CPU. Export to ONNX or TensorRT available to maximize FPS for your use case.

