Skip to content

Latest commit

 

History

History
128 lines (88 loc) · 4.22 KB

READMECH.md

File metadata and controls

128 lines (88 loc) · 4.22 KB

YOLODS

基于yolov5最新v6.0分支修改的多任务网络,可同时进行目标检测和分割.

Multi-task yolov5 with detection and segmentation
  • decoupled head
  • anchor free
  • segmentation head

Ablation experiment

在一个3k+训练1.5k+验证的47类数据集上实验:

model P R [email protected] [email protected]:95
yolov5s 0.536 0.368 0.374 0.206
yolov5s+train scrach 0.452 0.314 0.306 0.152
yolov5s+decoupled head 0.555 0.375 0.387 0.214
yolov5s + decoupled head+class balance weights 0.541 0.392 0.396 0.217
yolov5s + decoupled head+class balance weights 0.574 0.386 0.403 0.22
yolov5s + decoupled head+seghead 0.533 0.383 0.396 0.212

baseline模型为yolov5s.解耦头,类别权重等都可以提升MAP,添加分割头对检测效果影响甚微.

Training Method

python trainds.py

因为VOC数据集并没有同时提供所有的检测和分割标注，改为一个batch的检测和一个batch的分割数据集迭代训练,也即是训练时,一个batch检测和分割数据进行bp,然后梯度累积,然后统一进行更新参数.

MAP

这里为了和SSD512对比,采用yolov5s模型,输入为512:

model VOC2007 test
SSD512 79.8
yolov5s+seghead(512) 79.2

demo

see detectds.py. VOC上权重见weights:

Train custom data

  1. 使用labelme等工具标注检测框和分割mask;

    对于voc格式的box标注，可以使用 voc2yolo.py转为yolo格式；

    分割标注文件是json文件的话，需要转为类似VOC2012分割标签那种png格式。

  2. 默认已经熟悉如何训练yolov5，否则参考 how to arrange your detection dataset with yolov5 , 对于训练自己的数据集，可以参考VOC数据的训练data/voc.yaml:

    # Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]
path: .  # dataset root dir
train: VOC/det/images/train  # train images (relative to 'path') 118287 images
val: VOC/det/images/test  # train images (relative to 'path') 5000 images
road_seg_train: VOC/seg/images/train   # road segmentation data
road_seg_val: VOC/seg/images/val

# Classes
nc: 20  # number of classes
segnc: 20

names: ['aeroplane', 'bicycle', 'bird', 'boat',
           'bottle', 'bus', 'car', 'cat', 'chair',
           'cow', 'diningtable', 'dog', 'horse',
           'motorbike', 'person', 'pottedplant',
           'sheep', 'sofa', 'train', 'tvmonitor']  # class names

segnames: ['aeroplane', 'bicycle', 'bird', 'boat',
           'bottle', 'bus', 'car', 'cat', 'chair',
           'cow', 'diningtable', 'dog', 'horse',
           'motorbike', 'person', 'pottedplant',
           'sheep', 'sofa', 'train', 'tvmonitor']
    1. 修改trainds.py 中的参数：
    python trainds.py

    1. 如下测试图片文件夹：

      python detectds.py

Blog

  1. https://blog.csdn.net/IEEE_FELLOW/article/details/121912670
  2. https://blog.csdn.net/qq_57076285/article/details/124265887
  3. https://blog.csdn.net/sadjhaksdas/article/details/125762260

Reference

  1. YOLOP: You Only Look Once for Panoptic Driving Perception
  2. yolov5