Browse categories
Explore
Fiverr Pro
English
$
USD
I will build object detection, classification and pose estimation models
From lines of code to smiles!
Hello. My name is Arsenii, and I am a CS student at UCU. I am here to gain experience and help people who need my skills.
Another of my goals is to provide value. To achieve this, I do my job well an...
Level 1
Has met certain performance criteria and shows strong potential in the marketplace.
About this Gig
I offer professional custom models for object detection, classification, pose estimation, and segmentation. Models are built using top-tier architectures like YOLOv8, YOLOv11, Mask R-CNN, Swin Transformer, and ResNet.
I have completed numerous projects in medical imaging, industrial automation, quality assurance (defect detection), and human biomechanics (movement analysis).
The pipeline includes dataset preparation, training, evaluation (mAP, precision, recall, confusion matrix), and deployment-ready inference scripts. Real-time video processing is supported. Advanced ensemble models are used for critical accuracy tasks in premium solutions.
CPU, GPU, multi-GPU, and TPU training are supported. TPU accelerates training significantly for large datasets using TensorFlow, PyTorch/XLA, and HuggingFace Transformers (for Swin models).
All packages include a detailed evaluation report and a short testing period to ensure the model meets your accuracy and speed requirements.
APIs:
Google Cloud Vision API
Programming language:
Python
•
SQL
•
Colab
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
CVAT
•
Colab
•
PyTorch
Frameworks:
Scikit-learn
•
DeepPy
•
Keras
•
PyTorch
•
Panda
My Portfolio
FAQ
How many images do I need for good results?
Minimum 100 images per class for simple objects, 300 and more for complex scenarios. More data always improves performance, but I use advanced augmentation techniques to maximize results from smaller datasets.
What about ensemble methods in the Premium package?
Ensemble combines predictions from multiple models (e.g., YOLOv8, EfficientDet, Faster R-CNN) using techniques like Non-Maximum Suppression fusion or weighted averaging. Typically improves mAP by 2-5% but increases inference time.
Do you provide deployment support?
I deliver models in multiple formats (PyTorch, ONNX, TensorRT) with inference scripts. Basic deployment guidance included, but complex cloud/edge deployment is an add-on service.
