imatge-upc / salbow Goto Github PK

There are some errors when I running the code:

1. src/compute_assignments.py : extract_raw_features should be imported
2. src/extract_raw_features.py : os should be imported
3. the default backend of keras is tensorflow, while the code used theano
4. QE error: src/BLCF/utils_BLCF.py line182~184 miss else, it should be

if new_queries is None:
    new_queries = new
else:
    new_queries = vstack( (new_queries, new) )

When I run the evaluation.py, an error occurred. Does anyone know how to sovle it?

Using TensorFlow backend.
ImportError: No module named _multiarray_umath
terminate called after throwing an instance of 'std::runtime_error'
what(): numpy failed to initialize
Aborted

@evamohe Dear Eva, BLCF is efficient in terms of storage and computation. But if the dataset is huge like 100K 0r 1M images, do we still traverse the whole dataset to rank them all. How about the inverted index or hash code or something. @xavigiro

The code can run well if we have the precomputed data. But meet an error without the precomputed data.

Traceback (most recent call last):
File "evaluation.py", line 74, in
main()
File "/home/luliu/salbow/local/lib/python2.7/site-packages/click/core.py", line 722, in call
return self.main(*args, **kwargs)
File "/home/luliu/salbow/local/lib/python2.7/site-packages/click/core.py", line 697, in main
rv = self.invoke(ctx)
File "/home/luliu/salbow/local/lib/python2.7/site-packages/click/core.py", line 895, in invoke
return ctx.invoke(self.callback, **ctx.params)
File "/home/luliu/salbow/local/lib/python2.7/site-packages/click/core.py", line 535, in invoke
return callback(*args, **kwargs)
File "evaluation.py", line 45, in main
path_assignments = compute_assignments( ds, PATH_OUTPUT, layer, max_dim, mode='keyframes', interpolate=2 )
File "/home/luliu/work/salbow-master/src/compute_assignments.py", line 35, in compute_assignments
path_raw_features = extract_raw_features( ds, path_data, layer, max_dim, mode='keyframes' )
File "/home/luliu/work/salbow-master/src/extract_raw_features.py", line 22, in extract_raw_features
path_raw_features = compute_features( list_images, path_raw_features, layer, max_dim )
File "/home/luliu/work/salbow-master/src/feature_extraction/compute_features.py", line 24, in compute_features
feats = model.predict( preprocess_image( keyframe, max_dim=max_dim ) ).squeeze(axis=0)
File "/home/luliu/salbow/local/lib/python2.7/site-packages/keras/engine/training.py", line 1254, in predict
check_batch_axis=False)
File "/home/luliu/salbow/local/lib/python2.7/site-packages/keras/engine/training.py", line 124, in standardize_input_data
str(array.shape))
ValueError: Error when checking : expected input_1 to have shape (None, None, None, 3) but got array with shape (1, 3, 204, 340)

@evamohe

query_expansion in utils_BLCF.py
It seems that you've omitted an 'else:' before 'new_queries = vstack( (new_queries, new) )'

hi, I did not find the saliency net weighting in output.zip,could you post another URL