制作类似pascal voc格式的目标检测数据集:https://www.cnblogs.com/xiximayou/p/12546061.html

训练自己创建的数据集:https://www.cnblogs.com/xiximayou/p/12546556.html

验证自己创建的数据集:https://www.cnblogs.com/xiximayou/p/12550471.html

 

直接看修改后的text.py:

from __future__ import print_function
import sys
import os
import argparse
import torch
import torch.nn as nn
import torch.backends.cudnn as cudnn
import torchvision.transforms as transforms
from torch.autograd import Variable
#from data import VOC_ROOT, VOC_CLASSES as labelmap
from data import MASK_ROOT, MASK_CLASSES as labelmap
from PIL import Image
#from data import VOCAnnotationTransform, VOCDetection, BaseTransform, VOC_CLASSES
from data import MASKAnnotationTransform, MASKDetection, BaseTransform, MASK_CLASSES
import torch.utils.data as data
from ssd import build_ssd

parser = argparse.ArgumentParser(description='Single Shot MultiBox Detection')
parser.add_argument('--trained_model', default='weights/ssd300_MASK_5000.pth',
                    type=str, help='Trained state_dict file path to open')
parser.add_argument('--save_folder', default='eval/', type=str,
                    help='Dir to save results')
parser.add_argument('--visual_threshold', default=0.6, type=float,
                    help='Final confidence threshold')
parser.add_argument('--cuda', default=True, type=bool,
                    help='Use cuda to train model')
#parser.add_argument('--voc_root', default=VOC_ROOT, help='Location of VOC root directory')
parser.add_argument('-f', default=None, type=str, help="Dummy arg so we can load in Jupyter Notebooks")
args = parser.parse_args()

if args.cuda and torch.cuda.is_available():
    torch.set_default_tensor_type('torch.cuda.FloatTensor')
else:
    torch.set_default_tensor_type('torch.FloatTensor')

if not os.path.exists(args.save_folder):
    os.mkdir(args.save_folder)


def test_net(save_folder, net, cuda, testset, transform, thresh):
    # dump predictions and assoc. ground truth to text file for now
    filename = save_folder+'test1.txt'
    num_images = len(testset)
    for i in range(num_images):
        print('Testing image {:d}/{:d}....'.format(i+1, num_images))
        img = testset.pull_image(i)
        img_id, annotation = testset.pull_anno(i)
        x = torch.from_numpy(transform(img)[0]).permute(2, 0, 1)
        x = Variable(x.unsqueeze(0))

        with open(filename, mode='a') as f:
            f.write('nGROUND TRUTH FOR: '+img_id+'n')
            for box in annotation:
                f.write('label: '+' || '.join(str(b) for b in box)+'n')
        if cuda:
            x = x.cuda()

        y = net(x)      # forward pass
        detections = y.data
        # scale each detection back up to the image
        scale = torch.Tensor([img.shape[1], img.shape[0],
                             img.shape[1], img.shape[0]])
        pred_num = 0
        for i in range(detections.size(1)):
            j = 0
            while detections[0, i, j, 0] >= 0.6:
                if pred_num == 0:
                    with open(filename, mode='a') as f:
                        f.write('PREDICTIONS: '+'n')
                score = detections[0, i, j, 0]
                label_name = labelmap[i-1]
                pt = (detections[0, i, j, 1:]*scale).cpu().numpy()
                coords = (pt[0], pt[1], pt[2], pt[3])
                pred_num += 1
                with open(filename, mode='a') as f:
                    f.write(str(pred_num)+' label: '+label_name+' score: ' +
                            str(score) + ' '+' || '.join(str(c) for c in coords) + 'n')
                j += 1


def test_voc():
    # load net
    num_classes = len(MASK_CLASSES) + 1 # +1 background
    net = build_ssd('test', 300, num_classes) # initialize SSD
    net.load_state_dict(torch.load(args.trained_model))
    net.eval()
    print('Finished loading model!')
    # load data
    mask_root="/content/drive/My Drive/pytorch_ssd"
    testset = MASKDetection(mask_root, "test", None, MASKAnnotationTransform())
    if args.cuda:
        net = net.cuda()
        cudnn.benchmark = True
    # evaluation
    test_net(args.save_folder, net, args.cuda, testset,
             BaseTransform(net.size, (104, 117, 123)),
             thresh=args.visual_threshold)

if __name__ == '__main__':
    test_voc()

开始执行:

!python test.py --trained_model weights/ssd300_MASK_5000.pth

运行结果:

Finished loading model!
Testing image 1/80....
/pytorch/torch/csrc/autograd/python_function.cpp:622: UserWarning: Legacy autograd function with non-static forward method is deprecated and will be removed in 1.3. Please use new-style autograd function with static forward method. (Example: https://pytorch.org/docs/stable/autograd.html#torch.autograd.Function)
Testing image 2/80....
。。。。。。
/pytorch/torch/csrc/autograd/python_function.cpp:648: UserWarning: Legacy autograd function object was called twice.  You will probably get incorrect gradients from this computation, as the saved tensors from the second invocation will clobber the saved tensors from the first invocation.  Please consider rewriting your autograd function in the modern style; for information on the new format, please see: https://pytorch.org/docs/stable/notes/extending.html#extending-torch-autograd
Testing image 80/80....

看下生成了的文件:

【pytorch-ssd目标检测】测试自己创建的数据集

 

看下test1.py中是什么:

GROUND TRUTH FOR: test_00000007
label: 46.0 || 0.0 || 139.0 || 128.0 || 0
PREDICTIONS: 
1 label: mask score: tensor(0.9097) 31.465145 || 5.5611525 || 149.25903 || 86.10434

GROUND TRUTH FOR: test_00000010
label: 24.0 || 9.0 || 113.0 || 133.0 || 0
PREDICTIONS: 
1 label: mask score: tensor(0.8791) 21.426735 || 17.9471 || 112.9484 || 122.676765

GROUND TRUTH FOR: test_00000015
label: 407.0 || 37.0 || 486.0 || 143.0 || 0
PREDICTIONS: 
1 label: mask score: tensor(0.8441) 403.54123 || 42.476467 || 487.46075 || 146.36295

GROUND TRUTH FOR: test_00000016
label: 156.0 || 135.0 || 277.0 || 265.0 || 0
PREDICTIONS: 
1 label: mask score: tensor(0.9541) 159.74387 || 109.33117 || 284.67053 || 264.61325
。。。。。。

每一张图片的坐标、置信度。