Git Product home page Git Product logo

assn's Introduction

ASSN: Attention-based Scale Sequence Network

This repository contains the sources of ASSN, which improved the small object dectection performance. This includes the experimental results and the trained models on YOLOv7 and YOLOv8 based on ssFPN, a previous study conducted on YOLOv4 and YOLOR.

This repository was created based on the official repository source of YOLOv7 and YOLOv8.


Performance (MS-COCO)

Validation Set (5k images)

Model Test Size AP AP50 AP75 APS APM APL
YOLOv7 640 51.2% 69.7% 55.5% 35.2% 56.0% 66.7%
YOLOv7-X 640 52.9% 71.1% 57.5% 36.9% 57.7% 68.6%
YOLOv7-W6 1280 54.6% 72.3% 59.5% 40.1% 59.0% 68.6%
YOLOv8n 640 37.4% 52.9% 40.3% 18.6% 41.0% 53.5%
YOLOv8s 640 44.9% 62.1% 48.3% 25.9% 49.9% 61.0%
ASSN + YOLOv7 640 51.5% 70.0% 56.1% 35.7% 56.3% 65.8%
ASSN + YOLOv7-X 640 53.5% 71.6% 58.2% 36.7% 58.3% 69.7%
ASSN + YOLOv7-W6 1280 55.0% 72.7% 60.1% 40.0% 59.5% 68.2%
ASSN + YOLOv8n 640 37.4% 53.2% 40.5% 20.5% 41.6% 51.6%
ASSN + YOLOv8s 640 45.0% 62.3% 49.1% 27.2% 50.3% 59.7%

Test Set (test-dev2017, 20k images)

Model Test Size AP AP50 AP75 APS APM APL
YOLOv7 640 51.4% 69.7% 55.9% 31.8% 55.5% 65.0%
YOLOv7-X 640 53.1% 71.2% 57.8% 33.8% 57.1% 67.4%
YOLOv7-W6 1280 54.9% 72.6% 60.1% 37.3% 58.7% 67.1%
ASSN + YOLOv7 640 51.9% 70.3% 56.5% 32.6% 55.7% 65.4%
ASSN + YOLOv7-X 640 53.5% 71.5% 58.2% 34.3% 57.5% 67.4%
ASSN + YOLOv7-W6 1280 55.2% 72.9% 60.5% 37.9% 58.8% 67.5%

Testing Environment

CPU Intel® Core™ i9-9900K CPU @ 3.60GHz × 16
GPU NVIDIA GeForce GTX 1080 Ti (11GB)
OS Ubuntu 22.04.3 LTS

ASSN Architecture

architecture


Comparison between the ssFPN and proposed ASSN

comparison


assn's People

Contributors

yvvyee avatar

Stargazers

 avatar FrankHe avatar Byung-Gyu Kim avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.