dotayolov5s_improve

DOTAyolov5s_improve

https://github.com/daamachinelearning/dotayolov5s_improve

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 (3.2%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

DOTAyolov5s_improve

Basic Info
  • Host: GitHub
  • Owner: DaaMachineLearning
  • License: agpl-3.0
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 1.05 MB
Statistics
  • Stars: 2
  • Watchers: 1
  • Forks: 1
  • Open Issues: 0
  • Releases: 0
Created over 2 years ago · Last pushed over 2 years ago
Metadata Files
Readme Contributing License Citation

README.md

训练指令

datesets

将下载好的数据集放在文件夹(DOTAyolov5s_improve-main)中,在终端输入运行指令即可运行。

Put the downloaded dataset in the folder (DOTAyolov5s_improve-main) and enter the run command in the terminal to run.

DOTA_split: 链接:https://pan.baidu.com/s/1dunBf9Ib5yNqbNJmszIq0A 提取码:54p2

原模型训练指令

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5s.yaml

改进模型训练指令(CBAM注意力机制+EIOU损失函数+CoordConv卷积)训练指令

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMCoordConv.yaml

损失函数 CIOU / SIOU / EIOU / WIOU

utils/loss.py utils/metrics.py

基础训练指令

python train.py --img 640 --batch 16 --epochs 5 --data ./DOTA/datasets/DOTA.yaml --cfg ./models/DOTAyolov5s.yaml

python train.py --img 640 --batch 16 --epochs 5 --data ./data/coco128.yaml --cfg ./models/yolov5s.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sC3ECA.yaml --weights ./runs/train/exp20/weights/best.pt --resume ./runs/train/exp20/weights/last.pt

python train.py --img 640 --batch 16 --epochs 1 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5s.yaml

python train.py --img 640 --batch 16 --epochs 5 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sC3CBAM.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./split/DOTA_split.yaml --cfg ./models/DOTAyolov5sC3CBAM.yaml

python detect.py --weights best.pt --source ./data/images

class names

names: ['small-vehicle', 'large-vehicle', 'plane', 'storage-tank', 'ship', 'harbor', 'ground-track-field', 'soccer-ball-field', 'tennis-court', 'swimming-pool', 'baseball-diamond', 'roundabout', 'basketball-court', 'bridge', 'helicopter', 'container-crane']

2023/7/26

注意力机制 SE / SimAM / ECA / CoordAtt / CBAM

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5s.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sSE.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sSimAM.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sECA.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCoordAtt.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAM.yaml

卷积 SAConv / DCNConv / DSConv / CoordConv

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMSAConv.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMDCNConv.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMDSConv.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMCoordConv.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCoordAttCoordConv.yaml

检测头 dyhead

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMdyhead.yaml

空间金字塔池化改进 SPP / SPPF / ASPP / RFB / SPPCSPC

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMSPPCSPC.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMSPPCSPC_group.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMASPP.yaml

python train.py --img 640 --batch 16 --epochs 100 --data ./DOTAsplit/DOTAsplit.yaml --cfg ./models/DOTAyolov5sCBAMBasicRFB.yaml

Owner

  • Login: DaaMachineLearning
  • 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
Last Year