YOLO car detection section 3.5 error messages

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

I have passed the tests on all the graded functions and received 100% from the auto grader, however I keep getting this error message when I run the prediction in section 3.5. Any ideas? Thanks.

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-16-43edc87fb9a8> in <module>
----> 1 out_scores, out_boxes, out_classes = predict("test.jpg")

<ipython-input-15-85846e4b09d5> in predict(image_file)
     20     yolo_outputs = yolo_head(yolo_model_outputs, anchors, len(class_names))
     21 
---> 22     out_scores, out_boxes, out_classes = yolo_eval(yolo_outputs, [image.size[1],  image.size[0]], 10, 0.3, 0.5)
     23 
     24     # Print predictions info

<ipython-input-10-93b38129060e> in yolo_eval(yolo_outputs, image_shape, max_boxes, score_threshold, iou_threshold)
     31 
     32     # Use one of the functions you've implemented to perform Score-filtering with a threshold of score_threshold (ā‰ˆ1 line)
---> 33     scores, boxes, classes = yolo_filter_boxes(boxes, box_confidence, box_class_probs, threshold = score_threshold)
     34 
     35     # Scale boxes back to original image shape.

<ipython-input-3-10adbeca2b38> in yolo_filter_boxes(boxes, box_confidence, box_class_probs, threshold)
     40     ## (ā‰ˆ 3 lines)
     41     scores = tf.boolean_mask(box_class_scores, filtering_mask, axis=None)
---> 42     boxes = tf.boolean_mask(boxes, filtering_mask, axis=None)
     43     classes = tf.boolean_mask(box_classes, filtering_mask, axis=None)
     44     ### END CODE HERE

/opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py in wrapper(*args, **kwargs)
    199     """Call target, and fall back on dispatchers if there is a TypeError."""
    200     try:
--> 201       return target(*args, **kwargs)
    202     except (TypeError, ValueError):
    203       # Note: convert_to_eager_tensor currently raises a ValueError, not a

/opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/array_ops.py in boolean_mask_v2(tensor, mask, axis, name)
   1801   ```
   1802   """
-> 1803   return boolean_mask(tensor, mask, name, axis)
   1804 
   1805 

/opt/conda/lib/python3.7/site-packages/tensorflow/python/util/dispatch.py in wrapper(*args, **kwargs)
    199     """Call target, and fall back on dispatchers if there is a TypeError."""
    200     try:
--> 201       return target(*args, **kwargs)
    202     except (TypeError, ValueError):
    203       # Note: convert_to_eager_tensor currently raises a ValueError, not a

/opt/conda/lib/python3.7/site-packages/tensorflow/python/ops/array_ops.py in boolean_mask(tensor, mask, name, axis)
   1728     if axis_value is not None:
   1729       axis = axis_value
-> 1730       shape_tensor[axis:axis + ndims_mask].assert_is_compatible_with(shape_mask)
   1731 
   1732     leading_size = gen_math_ops.prod(shape(tensor)[axis:axis + ndims_mask], [0])

/opt/conda/lib/python3.7/site-packages/tensorflow/python/framework/tensor_shape.py in assert_is_compatible_with(self, other)
   1132     """
   1133     if not self.is_compatible_with(other):
-> 1134       raise ValueError("Shapes %s and %s are incompatible" % (self, other))
   1135 
   1136   def most_specific_compatible_shape(self, other):

ValueError: Shapes (1, 19, 19, 5) and (1, 19, 19, 80) are incompatible
2

It does seem pretty surprising that your code passes all the tests but fails in this way. One thing to check before debugging further is that your notebook is in a consistent state. Please try this sequence:

Kernel -> Restart and Clear Output
Save
Cell -> Run All

Then carefully scan through all the cells and see if all the tests pass and let us know whether you still get that same error.

3

Thanks for the quick response! I tried this sequence and got the same error messages. Previous cells still passed. By ā€œSaveā€ I did ā€œFile ā†’ Save and Checkpointā€.

What is ā€œconvert_to_eager_tensorā€? I couldnā€™t find any obvious from a quick google search. It looked as if section 3.5 and the previous cells were parsing the same code in different ways?

4

I donā€™t see any references to convert_to_eager_tensor in my YOLO notebook. Where did you see that?

5

OMG I found it through adding some prints to debug. The problem is in my yolo_filter_boxes implementation. When I specified the last dimension in tf.math.argmax() I used axis=3, which worked with the shape of inputs at that time. But in section 3.5 yolo_outputs has another dimension (totally 5 dimensions). I fixed my code with axis=-1 and all is well :slight_smile:

1 Like
6

Nice work! Thatā€™s great news that you found the solution. Thanks for the explanation.

7

convert_to_eager_tensor was in the error messages I got, see what I posted above in the large block of error messages. I still donā€™t know what it means but it was a false chase after all.

8

Ah, ok, that means itā€™s some internal function within the guts of TensorFlow. TF is open source, so weā€™d need to delve into the TF source to understand more details. I have not tried looking at the source, so I donā€™t know if you have access to the version corresponding to the actual TF version we are running here, which is 2.3.0 from sometime in 2021.

But, as you say, itā€™s probably a wild goose chase since you have already found the solution.