Does this skill support both PyTorch and TensorFlow?

Yes, the CV Pipeline Builder is designed to generate end-to-end pipelines compatible with both the PyTorch and TensorFlow frameworks.

How does it handle data augmentation?

It includes a dedicated ImageAugmentation module for standard transforms like rotation and cropping, as well as advanced techniques like Mixup, Cutout, and learned AutoAugment policies.

Can I use custom models with this skill?

While the skill provides industry-standard defaults like ResNet and YOLO, the generated pipelines are highly extensible, allowing you to integrate custom CNN architectures or Vision Transformers.

Is this skill suitable for production environments?

Absolutely. It includes built-in commands for model evaluation and exporting to ONNX, ensuring your computer vision models are optimized for high-performance production deployment.

Computer Vision Pipeline Builder

Name: Computer Vision Pipeline Builder
Author: anton-abyzov

byanton-abyzov

•

データサイエンスとML

Builds production-ready machine learning pipelines for image classification, object detection, and semantic segmentation using PyTorch or TensorFlow.

概要

The CV Pipeline Builder is a specialized machine learning skill for Claude Code designed to streamline the development of end-to-end computer vision systems. It automates complex tasks such as image preprocessing, advanced data augmentation, and transfer learning using industry-standard architectures like ResNet, YOLO, and U-Net. By integrating directly with the SpecWeave framework, it ensures that every vision model is developed with rigorous specifications, documented experiments, and seamless deployment paths to ONNX, making it ideal for transitioning ML research into production-grade software.

主な機能

Integrated experiment tracking and model versioning within SpecWeave increments
Automated data augmentation strategies including Mixup, Cutout, and AutoAugment
End-to-end CV pipeline generation for classification, detection, and segmentation
Seamless deployment exports for production-ready ONNX models
Pre-configured transfer learning for ResNet, EfficientNet, and Vision Transformers
13 GitHub stars

ユースケース

Developing real-time object detection systems for security or robotics applications
Implementing automated visual quality control in manufacturing and industrial workflows
Building medical imaging tools using pixel-level semantic segmentation

概要

主な機能

Integrated experiment tracking and model versioning within SpecWeave increments
Automated data augmentation strategies including Mixup, Cutout, and AutoAugment
End-to-end CV pipeline generation for classification, detection, and segmentation
Seamless deployment exports for production-ready ONNX models
Pre-configured transfer learning for ResNet, EfficientNet, and Vision Transformers
13 GitHub stars

ユースケース

Developing real-time object detection systems for security or robotics applications
Implementing automated visual quality control in manufacturing and industrial workflows
Building medical imaging tools using pixel-level semantic segmentation