Data

Tools

Indexes

Applications

Experiments

Open Source Science

yolov5_salient_feature_guided_knowledge_distillation

https://github.com/abrarzahin247/yolov5_salient_feature_guided_knowledge_distillation

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (8.6%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

Basic Info
  • Host: GitHub
  • Owner: AbrarZahin247
  • License: agpl-3.0
  • Language: Python
  • Default Branch: master
  • Size: 14.6 MB
Statistics
  • Stars: 2
  • Watchers: 1
  • Forks: 0
  • Open Issues: 7
  • Releases: 0
Created almost 2 years ago · Last pushed 6 months ago
Metadata Files
Readme Contributing License Citation

README.md

YOLOv5 Salient Feature Guided Knowledge Distillation on a Lightweight YOLOv5n-attention-light Model

Introduction

This repository contains the implementation of a novel approach for enhancing the performance of YOLOv5 models through Salient Feature Guided Knowledge Distillation (SFGKD). We introduce a lightweight variant of the YOLOv5n model, named YOLOv5n-attention-light, which incorporates attention mechanisms to improve detection accuracy while maintaining efficiency.

Features

  • Salient Feature Guided Knowledge Distillation (SFGKD): A technique to transfer knowledge from a larger, more accurate teacher model to a smaller, lightweight student model.
  • YOLOv5n-attention-light: A new lightweight YOLOv5n model variant with integrated attention mechanisms to enhance feature extraction.
  • Efficient and Accurate: Achieves a balance between model size, inference speed, and detection accuracy.

Model Architecture

The YOLOv5n-attention-light model builds upon the YOLOv5n architecture with additional attention layers to focus on important features within the input images. The SFGKD method further refines the model by distilling critical knowledge from a pre-trained, larger teacher model.

Installation

To get started, clone this repository and install the required dependencies:

bash git clone https://github.com/AbrarZahin247/YOLOv5_Salient_Map_Guided_Knowledge_Distillation.git && cd YOLOv5_Salient_Map_Guided_Knowledge_Distillation && git checkout light-attention pip install -r requirements.txt

Results

The YOLOv5n-attention-light model, trained with SFGKD, demonstrates improved detection accuracy on various benchmark datasets while maintaining a compact size and fast inference speed.

Contributing

We welcome contributions to this project. Please feel free to submit issues, feature requests, or pull requests.

Owner

  • Name: Root247
  • Login: AbrarZahin247
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.2.0
preferred-citation:
  type: software
  message: If you use YOLOv5, please cite it as below.
  authors:
  - family-names: Jocher
    given-names: Glenn
    orcid: "https://orcid.org/0000-0001-5950-6979"
  title: "YOLOv5 by Ultralytics"
  version: 7.0
  doi: 10.5281/zenodo.3908559
  date-released: 2020-5-29
  license: AGPL-3.0
  url: "https://github.com/ultralytics/yolov5"

GitHub Events

Total
  • Watch event: 1
  • Push event: 37
  • Create event: 2
Last Year
  • Watch event: 1
  • Push event: 37
  • Create event: 2

Dependencies

.github/workflows/ci-testing.yml actions
  • actions/checkout v4 composite
  • actions/setup-python v4 composite
  • slackapi/slack-github-action v1.24.0 composite
.github/workflows/codeql-analysis.yml actions
  • actions/checkout v4 composite
  • github/codeql-action/analyze v2 composite
  • github/codeql-action/autobuild v2 composite
  • github/codeql-action/init v2 composite
.github/workflows/docker.yml actions
  • actions/checkout v4 composite
  • docker/build-push-action v5 composite
  • docker/login-action v3 composite
  • docker/setup-buildx-action v3 composite
  • docker/setup-qemu-action v3 composite
.github/workflows/greetings.yml actions
  • actions/first-interaction v1 composite
.github/workflows/links.yml actions
  • actions/checkout v4 composite
  • nick-invision/retry v2 composite
.github/workflows/stale.yml actions
  • actions/stale v8 composite
.github/workflows/translate-readme.yml actions
  • actions/checkout v4 composite
  • actions/setup-node v4 composite
  • dephraiim/translate-readme main composite
utils/docker/Dockerfile docker
  • pytorch/pytorch 2.0.0-cuda11.7-cudnn8-runtime build
utils/google_app_engine/Dockerfile docker
  • gcr.io/google-appengine/python latest build
requirements.txt pypi
  • Pillow >=10.0.1
  • PyYAML >=5.3.1
  • gitpython >=3.1.30
  • matplotlib >=3.3
  • numpy >=1.22.2
  • opencv-python >=4.1.1
  • pandas >=1.1.4
  • psutil *
  • requests >=2.23.0
  • scipy >=1.4.1
  • seaborn >=0.11.0
  • setuptools >=65.5.1
  • thop >=0.1.1
  • torchvision >=0.9.0
  • tqdm >=4.64.0
  • ultralytics >=8.0.147
utils/google_app_engine/additional_requirements.txt pypi
  • Flask ==2.3.2
  • gunicorn ==19.10.0
  • pip ==23.3
  • werkzeug >=3.0.1