track 3 test_loader 문의

[cycos83]

안녕하세요, track 3의 모델을 submit하면 main.py의 infer 함수의 첫번째 for 문 (for idx, (image, fid) in enumerate(tqdm(test_loader)):) 이 한번만 돌고 끝납니다. 그래서 모든 테스트 데이터가 아니라 한 배치만큼의 데이터에 대해서만 모델이 판단하고 끝나버립니다. test_loader 부분은 수정한 곳이 없는데, 무엇이 문제인가요?

[cycos83]

infer 함수 부분입니다!

[cycos83]

`def _infer(model, root_path, test_loader=None): """Local inference function for NSML infer.

Args:
    model: instance. Any model is available.
    root_path: string. Automatically set by NSML.
    test_loader: instance. Data loader is defined in `data_local_loader.py`.

Returns:
    predictions_str: list of string.
                     ['img_1,1,0,1,0,1,0,0,0', 'img_2,0,1,0,0,1,0,0,0', ...]
"""

use_cuda = torch.cuda.is_available()
use_float16 = False
threshold = 0.3
iou_threshold = 0.3

model.eval()

if test_loader is None:
    test_loader = data_loader(root=os.path.join(root_path))

list_of_fids = []
list_of_preds = []

for idx, (image, fid) in enumerate(tqdm(test_loader)):
    print(idx)
    image = image.cuda()

    if use_cuda:
        model = model.cuda()
    if use_float16:
        model = model.half()

    with torch.no_grad():
        features, regression, classification, anchors = model(image)

        regressBoxes = BBoxTransform()
        clipBoxes = ClipBoxes()

        out = postprocess(image,
                          anchors, regression, classification,
                          regressBoxes, clipBoxes,
                          threshold, iou_threshold)
        out_class = []
        for n in range(len(image)):
            temp_out_class = out[n]['class_ids']
            out_class.append(temp_out_class)

            tmp_list = [0, 0, 0, 0, 0, 0, 0, 0]

            for i in range(len(temp_out_class)):

                for j in range(len(tmp_list)):
                        if out_class[n][i] == j:
                            tmp_list[j] = 1

            list_of_preds.append(tmp_list)

    list_of_fids.extend(fid)

    ############################################################################
list_of_fids = np.array(list_of_fids)
list_of_preds = np.array(list_of_preds)

predictions_str = []

for idx, fid in enumerate(list_of_fids):
    test_str = fid

    for pred in list_of_preds[idx]:
        test_str += ',{}'.format(pred)

    predictions_str.append(test_str)

predictions_str = np.array(predictions_str)

return predictions_str`