The xInfer Model Zoo API

Welcome to the xInfer::zoo. This is the high-level, "batteries-included" API for xInfer.

The zoo is a collection of pre-packaged, hyper-optimized, and incredibly easy-to-use solutions for the world's most common AI tasks.

The Philosophy: Solutions, Not Just Tools

While the Core Toolkit provides the powerful, low-level "engine parts" for building custom pipelines, the zoo provides the finished "F1 car."

Each class in the zoo is a complete, end-to-end pipeline that abstracts away all the complexity of pre-processing, inference, and post-processing. The goal is to let you solve a complex problem like real-time object detection or image generation with just two lines of C++ code:

  1. One line to initialize the pipeline from a pre-built engine.
  2. One line to predict.

This is the power of the zoo. It gives you the full, state-of-the-art performance of a custom C++/CUDA/TensorRT application with the simplicity of a high-level library.

Key Features of All zoo Classes

  • Performance by Default: Every zoo class is built on top of the hyper-performant xInfer::core::InferenceEngine and uses fused CUDA kernels from the preproc and postproc modules wherever possible.
  • Simple, Task-Oriented API: You don't interact with raw tensors. You provide a cv::Mat image and get back a std::vector<BoundingBox> or a std::string. The API is designed around the final answer, not the intermediate steps.
  • Seamless Hub Integration: Most zoo classes have special constructors that can download a pre-built, perfectly optimized engine for your hardware directly from the Ignition Hub. This provides a "zero-setup" user experience.
  • Robust and Production-Ready: These classes are designed to be used directly in your final application. They are efficient, safe, and easy to integrate.

The zoo Catalog of Solutions

The zoo is organized by domain. Explore the available pipelines below to find the solution you need.

🖼️ Computer Vision

Tools for understanding and analyzing visual information from images and video. This is the most mature and comprehensive part of the zoo.

  • Tasks: Image Classification, Object Detection, Semantic & Instance Segmentation, Pose Estimation, Face Recognition, OCR, and many more.

➡️ Explore the Vision API

✨ Generative AI

Powerful pipelines for creating novel content, from images and audio to 3D models.

  • Tasks: Text-to-Image (Diffusion), Image Generation (GANs), Super-Resolution, Style Transfer, Text-to-Speech.

➡️ Explore the Generative API

📝 Natural Language Processing (NLP)

High-throughput, low-latency solutions for understanding and processing human language.

  • Tasks: Text Classification, Named Entity Recognition (NER), Sentence Embeddings for RAG, Summarization, and Translation.

➡️ Explore the NLP API

🎧 Audio & Signal Processing

Real-time pipelines for analyzing audio signals, from speech to environmental sounds.

  • Tasks: Speech Recognition, Audio Classification, Speaker Identification, Music Source Separation.

➡️ Explore the Audio & DSP API

📈 Time Series

Specialized solutions for forecasting and analyzing sequential data.

  • Tasks: Forecasting, Anomaly Detection, and Classification.

➡️ Explore the Time Series API

🧊 3D & Spatial Computing

Cutting-edge pipelines for processing 3D data from sensors like LIDAR.

  • Tasks: 3D Reconstruction (Gaussian Splatting), Point Cloud Detection, Point Cloud Segmentation.

➡️ Explore the 3D API

🌍 Geospatial

Specialized tools for analyzing satellite and aerial imagery.

  • Tasks: Building & Road Segmentation, Change Detection, Maritime Object Detection.

➡️ Explore the Geospatial API

⚕️ Medical Imaging

High-performance pipelines for medical image analysis.

  • Tasks: Tumor Detection, Cell Segmentation, Retinal Abnormality Scanning, Artery Analysis.

➡️ Explore the Medical API

📄 Document AI

Pipelines for understanding the structure and content of documents.

  • Tasks: Table Extraction, Signature Detection, Handwriting Recognition.

➡️ Explore the Document AI API

🚀 Specialized & RL

Hyper-specialized, high-value solutions for specific industries and advanced applications.

  • Tasks: Reinforcement Learning Policies, Financial (HFT) Models, Physics Simulation, Genomics.

➡️ Explore the Specialized API