# wang-xinyu/tensorrtx

Implementation of popular deep learning networks with TensorRT network definition API

Repository: https://github.com/wang-xinyu/tensorrtx
Canonical: https://ross.abutalabs.com/products/tensorrtx
Language: C++
License: MIT
License Family: permissive
Topics: tensorrt, mnasnet, resnet, squeezenet, yolov3, mobilenetv2, mobilenetv3, yolov3-spp, retinaface, yolov4, arcface, yolov5, crnn, swin-transformer, detr, yolov7, yolov8, yolov9, yolo11
Last push: 2026-08-18T07:11:41+00:00

## Health v2 (maintenance only)
Score: 76/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 35, longevity 100
- inputs: {"age_days": 2473, "days_push": 15, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 7827, forks 1858 (observed 2026-08-28T04:10:05.309721+00:00)

## What it is
A C++ collection of popular deep learning networks (YOLO variants, ResNet, MobileNet, Swin Transformer, OCR models, and more) implemented from scratch with the NVIDIA TensorRT network definition API. It exports trained weights to .wts files and builds optimized TensorRT inference engines without relying on ONNX or other parsers.

## Use cases
- deploy yolo models with tensorrt in c++
- build tensorrt engines without onnx parser
- run deep learning inference on nvidia gpu with maximum performance
- convert pytorch weights to tensorrt
- learn how popular networks are structured layer by layer
- integrate preprocessing and postprocessing into a tensorrt network
- deploy vision transformer or swin transformer on jetson

## When to choose
- you need fine-grained control over TensorRT network layers or want to modify/merge layers
- ONNX or UFF parsers fail or produce suboptimal engines for your model
- you want to learn network architectures by building them explicitly
- you need a reference C++ implementation of YOLO or other vision models on TensorRT

## When to avoid
- your model exports cleanly to ONNX and trtexec or onnx-tensorrt works fine
- you want a high-level Python deployment pipeline with minimal code
- you need training or fine-tuning - this is inference-only
- you target non-NVIDIA hardware

## Facets
- artifact type: library
- maturity: active
- function: deep-learning, machine-learning, computer-vision, image-processing, ocr, llm-inference
- domain: deep-learning, computer-vision, machine-learning, gpu-computing
- platform: cpp, windows, cross-platform
- tags: tensorrt, nvidia, inference, yolo, model-deployment, network-definition-api, wts-weights, engine-building, gpu, linux

## Member repositories
- wang-xinyu/tensorrtx (main) score 76

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:05.309721+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:34:08.496747+00:00, confidence not recorded.
  - readme: https://github.com/wang-xinyu/tensorrtx (fetched 2026-08-28T04:10:05.309721+00:00, sha a0b3abdd159c)
- Data as of 2026-08-30T08:39:29.467469+00:00.