# PengyiZhang/SlimYOLOv3

This page is for the SlimYOLOv3: Narrower, Faster and Better for UAV Real-Time Applications

Repository: https://github.com/PengyiZhang/SlimYOLOv3
Canonical: https://ross.abutalabs.com/products/slimyolov3
Language: Python
License Family: other
Last push: 2019-10-25T06:05:33+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2596, "days_push": 2504, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1152, forks 315 (observed 2026-08-28T04:03:47.210048+00:00)

## What it is
SlimYOLOv3 is a research implementation of channel-pruned YOLOv3 object detectors optimized for real-time inference on UAVs and embedded devices. It provides a training pipeline (normal training, L1 sparsity training, channel pruning, fine-tuning) built on PyTorch and darknet, evaluated on the VisDrone2018-Det benchmark.

## Use cases
- run real-time object detection on drones with limited compute
- prune YOLOv3 channels to reduce FLOPs and parameters
- train a sparse object detector with L1 regularization on channel scaling factors
- fine-tune a pruned YOLOv3 model to recover accuracy
- evaluate pruned YOLOv3 models on the VisDrone dataset
- deploy lightweight object detection on embedded devices

## When to choose
- you need a faster, smaller YOLOv3 for UAV or embedded real-time detection
- you want to apply channel pruning to YOLOv3 with a proven recipe
- you are working with the VisDrone aerial detection benchmark

## When to avoid
- you need a maintained, production-ready detection framework (last release 2019, no license)
- you need detection on general images rather than aerial/UAV scenes
- you want plug-and-play pretrained models without running the pruning pipeline yourself
- you require a permissively licensed project for commercial use

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, computer-vision, image-processing, deep-learning
- domain: computer-vision, machine-learning, deep-learning
- platform: python
- tags: object-detection, yolo, channel-pruning, uav, drone, model-compression, pytorch, darknet, real-time, linux, gpu

## Member repositories
- PengyiZhang/SlimYOLOv3 (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:47.210048+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 2026-08-30T06:33:31.552623+00:00, confidence not recorded.
  - readme: https://github.com/PengyiZhang/SlimYOLOv3 (fetched 2026-08-28T04:03:47.210048+00:00, sha b3f948c3b2df)
- Data as of 2026-08-30T08:39:29.467469+00:00.