模型实战一之YOLOv7实例分割、模型训练自己数据集

1.环境准备

• 下载yolov7实例分割模型：

git clone https://github.com/WongKinYiu/yolov7.git -b mask yolov7-mask

cd yolov7-mask

• 安装环境

#查看已安装环境
conda info --envs
#查看安装了哪些包
conda list

#创建环境 
conda create -n yolov7 python=3.8
#激活
conda activate yolov7

# 安装 torch 1.8.2+cu11.1
pip install torch==1.8.2 torchvision==0.9.2 torchaudio===0.8.2 --extra-index-url https://download.pytorch.org/whl/lts/1.8/cu111

#其他版本：torch+cuda10.2
pip install torch==1.8.1+cu102 torchvision==0.9.1+cu102 torchaudio===0.8.1 -f https://download.pytorch.org/whl/torch_stable.html


# 修改requirements.txt，将其中的torch和torchvision注释掉
pip install -r requirements.txt

• 安装detectron2
  detectron是facebook发布的开源机器视觉库，安装教程参考：https://blog.csdn.net/qq_45770232/article/details/126471738

# 安装detectron2
#先安装ninja
pip install ninja

git clone https://github.com/facebookresearch/detectron2
cd detectron2
python setup.py install
cd ..

2.测试实例分割demo

• 测试：

下载权重放在detect.py路径下：yolov7.pt ... yolov7-mask.pt

测试yolov7目标检测：

 python detect.py --weights yolov7.pt --conf 0.25 --img-size 640 --source inference/images/horses.jpg 

• 测试实例分割 - python

import matplotlib.pyplot as plt
import torch
import cv2
import yaml
from torchvision import transforms
import numpy as np

from utils.datasets import letterbox
from utils.general import non_max_suppression_mask_conf

from detectron2.modeling.poolers import ROIPooler
from detectron2.structures import Boxes
from detectron2.utils.memory import retry_if_cuda_oom
from detectron2.layers import paste_masks_in_image

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
with open('data/hyp.scratch.mask.yaml') as f:
    hyp = yaml.load(f, Loader=yaml.FullLoader)
weigths = torch.load('./weights/yolov7-mask.pt')
model = weigths['model']
model = model.half().to(device)
_ = model.eval()

image = cv2.imread('inference/images/horses.jpg')  # 504x378 image
image = letterbox(image, 640, stride=64, auto=True)[0]
image_ = image.copy()
image = transforms.ToTensor()(image)
image = torch.tensor(np.array([image.numpy()]))
image = image.to(device)
image = image.half()

output = model(image)

inf_out, train_out, attn, mask_iou, bases, sem_output = output['test'], output['bbox_and_cls'], output['attn'], output['mask_iou'], output['bases'], output['sem']

bases = torch.cat([bases, sem_output], dim=1)
nb, _, height, width = image.shape
names = model.names
pooler_scale = model.pooler_scale
pooler = ROIPooler(output_size=hyp['mask_resolution'], scales=(pooler_scale,), sampling_ratio=1, pooler_type='ROIAlignV2', canonical_level=2)

output, output_mask, output_mask_score, output_ac, output_ab = non_max_suppression_mask_conf(inf_out, attn, bases, pooler, hyp, conf_thres=0.25, iou_thres=0.65, merge=False, mask_iou=None)

pred, pred_masks = output[0], output_mask[0]
base = bases[0]
bboxes = Boxes(pred[:, :4])
original_pred_masks = pred_masks.view(-1, hyp['mask_resolution'], hyp['mask_resolution'])
pred_masks = retry_if_cuda_oom(paste_masks_in_image)( original_pred_masks, bboxes, (height, width), threshold=0.5)
pred_masks_np = pred_masks.detach().cpu().numpy()
pred_cls = pred[:, 5].detach().cpu().numpy()
pred_conf = pred[:, 4].detach().cpu().numpy()
nimg = image[0].permute(1, 2, 0) * 255
nimg = nimg.cpu().numpy().astype(np.uint8)
nimg = cv2.cvtColor(nimg, cv2.COLOR_RGB2BGR)
nbboxes = bboxes.tensor.detach().cpu().numpy().astype(np.int32)
pnimg = nimg.copy()

for one_mask, bbox, cls, conf in zip(pred_masks_np, nbboxes, pred_cls, pred_conf):
    if conf < 0.25:
        continue
    color = [np.random.randint(255), np.random.randint(255), np.random.randint(255)]

    pnimg[one_mask] = pnimg[one_mask] * 0.5 + np.array(color, dtype=np.uint8) * 0.5
    pnimg = cv2.rectangle(pnimg, (bbox[0], bbox[1]), (bbox[2], bbox[3]), color, 2)
    #label = '%s %.3f' % (names[int(cls)], conf)
    #t_size = cv2.getTextSize(label, 0, fontScale=0.5, thickness=1)[0]
    #c2 = bbox[0] + t_size[0], bbox[1] - t_size[1] - 3
    #pnimg = cv2.rectangle(pnimg, (bbox[0], bbox[1]), c2, color, -1, cv2.LINE_AA)  # filled
    #pnimg = cv2.putText(pnimg, label, (bbox[0], bbox[1] - 2), 0, 0.5, [255, 255, 255], thickness=1, lineType=cv2.LINE_AA)


# coco example
# %matplotlib inline
cv2.imshow("instance_result.jpg", pnimg)
cv2.waitKey(0)
# cv2.imwrite("instance_result.jpg", pnimg)

3.训练自己的数据集

• 实例分割时目标检测与语义分割的结合，所以其标注文件初始为通过labelme标注的json格式，要用yolo模型进行训练，需要将其转换为yolo所需要的txt格式：
  
• 转换demo如下：
  参考：https://blog.csdn.net/qq_57329395/article/details/128079776

# 处理labelme多边形矩阵的标注  json转化txt
import json
import os

name2id = {'peanuthull': 0, 'kernel': 1}


def convert(img_size, box):
    dw = 1. / (img_size[0])
    dh = 1. / (img_size[1])
    x = (box[0] + box[2]) / 2.0
    y = (box[1] + box[3]) / 2.0
    w = abs(box[2] - box[0])
    h = abs(box[3] - box[1])
    x = x * dw
    w = w * dw
    y = y * dh
    h = h * dh
    return (x, y, w, h)


def decode_json(json_floder_path, txt_outer_path, json_name):
    #  json_floder_path='E:\\Python_package\\itesjson\\'
    # json_name='V1125.json'
    txt_name = txt_outer_path + json_name[:-5] + '.txt'
    with open(txt_name, 'w') as f:
        json_path = os.path.join(json_floder_path, json_name)  # os路径融合
        data = json.load(open(json_path, 'r', encoding='gb2312', errors='ignore'))
        img_w = data['imageWidth']  # 图片的高
        img_h = data['imageHeight']  # 图片的宽
        isshape_type = data['shapes'][0]['shape_type']
        print(isshape_type)
        # print(isshape_type)
        # print('下方判断根据这里的值可以设置为你自己的类型，我这里是polygon'多边形)
        # len(data['shapes'])
        for i in data['shapes']:
            label_name = i['label']  # 得到json中你标记的类名
            if (i['shape_type'] == 'polygon'):  # 数据类型为多边形 需要转化为矩形
                x_max = 0
                y_max = 0
                x_min = 100000
                y_min = 100000
                for lk in range(len(i['points'])):
                    x1 = float(i['points'][lk][0])
                    y1 = float(i['points'][lk][1])
                    # print(x1)
                    if x_max < x1:
                        x_max = x1
                    if y_max < y1:
                        y_max = y1
                    if y_min > y1:
                        y_min = y1
                    if x_min > x1:
                        x_min = x1
                bb = (x_min, y_max, x_max, y_min)
            if (i['shape_type'] == 'rectangle'):  # 为矩形不需要转换
                x1 = float(i['points'][0][0])
                y1 = float(i['points'][0][1])
                x2 = float(i['points'][1][0])
                y2 = float(i['points'][1][1])
                bb = (x1, y1, x2, y2)
            bbox = convert((img_w, img_h), bb)
            try:
                f.write(str(name2id[label_name]) + " " + " ".join([str(a) for a in bbox]) + '\n')
            except:
                pass


if __name__ == "__main__":
    json_floder_path = 'data_\\jsons\\'  # 存放json的文件夹的绝对路径
    txt_outer_path = 'data_\\txts\\'  # 存放txt的文件夹绝对路径
    json_names = os.listdir(json_floder_path)
    print("共有：{}个文件待转化".format(len(json_names)))
    flagcount = 0
    for json_name in json_names:
        decode_json(json_floder_path, txt_outer_path, json_name)
        flagcount += 1
        print("还剩下{}个文件未转化".format(len(json_names) - flagcount))

    # break
    print('转化全部完毕')

• 数据集存放格式：

• datasets:

  • images:

    • train: .jpg
    • val: .jpg

  • labels:

    • train: .txt
    • val: .txt

  • train_list.txt

  • val_list.txt

• train_list 及val_list存放绝对路径，如下：
  
  

参考：https://blog.csdn.net/matt45m/article/details/127416919?spm=1001.2014.3001.5502