InvalidArgumentError: ConcatOp : Dimensions of inputs should match: shape[0] = [2704,54] vs. shape[1] = [8112,18] [Op:ConcatV2] name: concat

[VedVyapak]

I have changed the test file according to my use in colab.

I am getting the following error:

InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-90-cc7e8926b76f> in <module>()
     75     # X = np.array(pred_bbox, dtype="object")
     76     # print(X.shape)
---> 77     pred_bbox = tf.concat(pred_bbox, axis=0)
     78     bboxes = postprocess_boxes(pred_bbox, image_size, INPUT_SIZE, SCORE_THRESHOLD)
     79     bboxes = nms(bboxes, IOU_THRESHOLD , method='nms')

4 frames
/usr/local/lib/python3.7/dist-packages/six.py in raise_from(value, from_value)

InvalidArgumentError: ConcatOp : Dimensions of inputs should match: shape[0] = [2704,54] vs. shape[1] = [8112,18] [Op:ConcatV2] name: concat

I have put print function many times in my code to understand the code and I have also mentioned the output in comments wherever the print function is present. My code is

predicted_dir_path = '/content/drive/MyDrive/Okutama/mAP/predicted'
ground_truth_dir_path = '/content/drive/MyDrive/Okutama/mAP/ground-truth'

INPUT_SIZE   = 416
NUM_CLASS    = 13
CLASSES      = ['"None"\n', '"Handshaking"\n', '"Hugging"\n', '"Reading"\n', '"Drinking"\n', '"Pushing/Pulling"\n', '"Carrying"\n', '"Calling"\n', '"Running"\n', '"Walking"\n', '"Lying"\n', '"Sitting"\n', '"Standing"\n']

feature_maps = YOLOv4(input_layer)

bbox_tensors = []
for i, fm in enumerate(feature_maps):
    if i == 0:
        bbox_tensor = decode_train(fm, 52, 13, 8, ANCHORS = anchors, i = i, XYSCALE = xyscale)
    elif i == 1:
        bbox_tensor = decode_train(fm, 26, 13, 16, ANCHORS = anchors, i = i, XYSCALE = xyscale)
    else:
        bbox_tensor = decode_train(fm, 13, 13, 32, ANCHORS = anchors, i = i, XYSCALE = xyscale)
    bbox_tensors.append(fm)
    bbox_tensors.append(bbox_tensor)

model = tf.keras.Model(input_layer, bbox_tensors)
model.load_weights("/content/drive/MyDrive/Okutama/YoloV4_2.h5")

detetction_path = '/content/drive/MyDrive/Okutama/detection'
IOU_THRESHOLD = 0.45 
SCORE_THRESHOLD = 0.3

for num in range(len(x_test)):
    image = cv2.imread(x_test[num])
    boxes = load_annotation_test(filename=y_test[num], frame=frame_no_test[num])
    # boxes = load_annotation(filename=y_test[i%len(y_test)], frame=frame_no_test[i%len(frame_no_test)])
    # bbox_data_gt = np.array([list(map(int, box.split(','))) for box in boxes])
    bbox_data_gt = np.array(boxes)

    if len(bbox_data_gt) == 0:
        bboxes_gt=[]
        classes_gt=[]
    else:
        bboxes_gt, classes_gt = bbox_data_gt[:, :4], bbox_data_gt[:, 4]

    ground_truth_path = os.path.join(ground_truth_dir_path, str(num) + '.txt')

    image_path = x_test[0]
    image_name = image_path.split('/')[-1]

    print('=> ground truth of %s:' % image_name)
    num_bbox_gt = len(bboxes_gt)

    with open(ground_truth_path, 'w') as f:
        for i in range(num_bbox_gt):
            class_name = CLASSES[int(classes_gt[i])]
            xmin, ymin, xmax, ymax = list(map(str, bboxes_gt[i]))
            bbox_mess = ' '.join([class_name, xmin, ymin, xmax, ymax]) + '\n'
            f.write(bbox_mess)
            print('\t' + str(bbox_mess).strip())

    print('=> predict result of %s:' % image_name)
    predict_result_path = os.path.join(predicted_dir_path, str(num) + '.txt')
    # Predict Process.

    print(image.shape)  # (720, 1280, 3)
    image_size = image.shape[:2]
    print(image_size)    # (720, 1280)
    image_data = image_preporcess(np.copy(image), [INPUT_SIZE, INPUT_SIZE])
    print(image_data.shape)  #(416, 416, 3)
    image_data = image_data[np.newaxis, ...].astype(np.float32)
    print(image_data.shape)  #(1, 416, 416, 3)

    pred_bbox = model.predict(image_data)
    pred_bbox = [tf.reshape(x, (-1, tf.shape(x)[-1])) for x in pred_bbox]
    # X = np.array(pred_bbox, dtype="object")
    # print(X.shape)
    pred_bbox = tf.concat(pred_bbox, axis=0)
    bboxes = postprocess_boxes(pred_bbox, image_size, INPUT_SIZE, SCORE_THRESHOLD)
    bboxes = nms(bboxes, IOU_THRESHOLD , method='nms')

[pythonlessons]

Cool that you could solve this by your self