# ppogg/YOLOv5-Lite

🍅🍅🍅YOLOv5-Lite: Evolved from yolov5 and the size of model is only 900+kb (int8) and 1.7M (fp16). Reach 15 FPS on the Raspberry Pi 4B~

Repository: https://github.com/ppogg/YOLOv5-Lite
Canonical: https://ross.abutalabs.com/products/yolov5-lite
Language: C++
License: GPL-3.0
License Family: copyleft
Topics: ncnn, yolov5, android-app, mnn, shufflenetv2, repvgg, transformer, tensorrt, pplcnet, openvivo, mobilenet, tflite, pytorch, picodet, onnxruntime
Last push: 2024-06-22T02:13:22+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1843, "days_push": 803, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2487, forks 430 (observed 2026-08-28T04:06:55.640688+00:00)

## What it is
YOLOv5-Lite is a lightweight object detection model family evolved from YOLOv5, with models as small as ~900KB (int8) that run 10-15+ FPS on a Raspberry Pi 4B. It includes training code in PyTorch plus deployment support for inference backends like ncnn, MNN, TensorRT, OpenVINO, and TFLite on edge devices.

## Use cases
- run object detection on a raspberry pi
- deploy a small yolo model on android
- detect objects in real time on embedded devices
- train a lightweight yolo model in pytorch
- convert a detection model to ncnn or mnn for mobile inference
- run yolo inference with tensorrt or openvino
- quantize a detection model to int8 for edge deployment

## When to choose
- you need real-time object detection on CPU-only or low-power hardware like Raspberry Pi or Android phones
- model size and memory footprint are critical constraints
- you want prebuilt export paths to ncnn, MNN, TensorRT, OpenVINO, or TFLite

## When to avoid
- you need maximum accuracy on high-end GPUs where full YOLOv5 or YOLOv8 is better
- you need segmentation, pose, or other vision tasks beyond bounding-box detection
- you require a permissive license since this is GPL-3.0

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, machine-learning, deep-learning, embedded-systems
- platform: cross-platform, cpp, python
- tags: object-detection, yolo, yolov5, ncnn, mnn, tensorrt, openvino, tflite, onnx, raspberry-pi, edge-deployment, model-compression, quantization, lightweight-model, shufflenet, linux, android

## Member repositories
- ppogg/YOLOv5-Lite (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:55.640688+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 2026-08-30T02:28:01.298886+00:00, confidence not recorded.
  - readme: https://github.com/ppogg/YOLOv5-Lite (fetched 2026-08-28T04:06:55.640688+00:00, sha 480436e3661f)
- Data as of 2026-08-30T08:39:29.467469+00:00.