def eval(model, save_path, test_path, device):
model.eval()
# torch.cuda.empty_cache() # speed up evaluating after training finished
img_path = os.path.join(test_path, 'img')
gt_path = os.path.join(test_path, 'gt')
if os.path.exists(save_path):
shutil.rmtree(save_path, ignore_errors=True)
if not os.path.exists(save_path):
os.makedirs(save_path)
long_size = 2240
# 预测所有测试图片
img_paths = [os.path.join(img_path, x) for x in os.listdir(img_path)]
for img_path in tqdm(img_paths, desc='test models'):
img_name = os.path.basename(img_path).split('.')[0]
save_name = os.path.join(save_path, 'res_' + img_name + '.txt')
assert os.path.exists(img_path), 'file is not exists'
img = cv2.imread(img_path)
h, w = img.shape[:2]
# if max(h, w) > long_size:
scale = long_size / max(h, w)
img = cv2.resize(img, None, fx=scale, fy=scale)
# 将图片由(w,h)变为(1,img_channel,h,w)
tensor = transforms.ToTensor()(img)
tensor = tensor.unsqueeze_(0)
tensor = tensor.to(device)
with torch.no_grad():
preds = model(tensor)
preds, boxes_list = pse_decode(preds[0], config.scale)
scale = (preds.shape[1] * 1.0 / w, preds.shape[0] * 1.0 / h)
if len(boxes_list):
boxes_list = boxes_list / scale
np.savetxt(save_name,
boxes_list.reshape(-1, 8),
delimiter=',',
fmt='%d')
# 开始计算 recall precision f1
result_dict = cal_recall_precison_f1(gt_path, save_path)
return result_dict['recall'], result_dict['precision'], result_dict[
'hmean']
cv2.imwrite(os.path.join(save_img_folder, '{}.jpg'.format(img_name)),
img)
np.savetxt(save_name,
boxes_list.reshape(-1, 8),
delimiter=',',
fmt='%d')
print('fps:{}'.format(total_frame / total_time))
return save_txt_folder
if __name__ == '__main__':
os.environ['CUDA_VISIBLE_DEVICES'] = str('0')
backbone = 'resnet101'
scale = 4
model_path = './output/APSEGAT_weights.pth'
data_path = './images/img'
gt_path = './images/gt'
save_path = './result/_scale{}'.format(scale)
gpu_id = 0
print('backbone:{},scale:{},model_path:{}'.format(backbone, scale,
model_path))
save_path = main(model_path,
backbone,
scale,
data_path,
save_path,
gpu_id=gpu_id)
result = cal_recall_precison_f1(gt_path=gt_path, result_path=save_path)
print(result)
# print(cal_recall_precison_f1('/data2/dataset/ICD151/test/gt', '/data1/zj/tensorflow_PSENet/tmp/'))
if not os.path.exists(img_save_path):
os.mkdir(img_save_path)
detect = OcrDetCTPN(opt.model_path)
image_files = os.listdir(opt.image_root)
for i in tqdm(range(len(image_files))):
image_file = image_files[i]
im_name = image_file.split('.')[0]
fid = open(os.path.join(txt_save_path, 'res_' + im_name + '.txt'),
'w+',
encoding='utf-8')
image_file = os.path.join(opt.image_root, image_file)
image = cv2.imread(image_file)
det_recs = detect.inference(image)
for i, box in enumerate(det_recs):
x3, y3 = box[6], box[7]
box[6], box[7] = box[4], box[5]
box[4], box[5] = x3, y3
box = box[:8].reshape(4, 2).astype(np.int32)
cv2.polylines(image, [box[:8].reshape((-1, 1, 2))],
True,
color=(0, 255, 0),
thickness=2)
box = [str(x) for x in box.reshape(-1).tolist()]
fid.write(','.join(box) + '\n')
cv2.imwrite(os.path.join(img_save_path, im_name + '.jpg'), image)
fid.close()
pre = cal_recall_precison_f1(opt.test_gt, txt_save_path, show_result=False)
print(pre)
if not os.path.exists(save_txt_folder):
os.makedirs(save_txt_folder)
start_time_all = time.time()
img_count = 0
for i in tqdm.tqdm(os.listdir(test_images_path)):
# print(i)
img_name = i.split('.')[0]
i = os.path.join(test_images_path, i)
img = cv2.imread(i)
start_time = time.time()
prediction, vis_output, polygons = detection_demo.run_on_image(img)
img_save_path = os.path.join(save_img_folder,
'res_' + img_name + '.jpg')
txt_save_path = os.path.join(save_txt_folder,
'res_' + img_name + '.txt')
save_result_to_txt(txt_save_path, prediction, polygons)
# print("Time: {:.2f} s / img".format(time.time() - start_time))
vis_output.save(img_save_path)
img_count += 1
print("Average Time: {:.2f} s /img".format(
(time.time() - start_time_all) / img_count))
result = cal_recall_precison_f1(gt_path=args.gt_path,
result_path=save_txt_folder)
print(result)
