face-search icon indicating copy to clipboard operation
face-search copied to clipboard

need some additional information

Open vaboston opened this issue 8 years ago • 4 comments

Your project is very interesting but I have a little trouble to understand how to use the run.sh , at what point you indicate the downloaded pictures ? Thanks

vaboston avatar Sep 16 '16 14:09 vaboston

Hi,

Please follow the instructions in README.md. Initially you will need to download images. Feel free to expand the demo dataset with new images. On big dataset you should get fantastic results. This is what I get on 250K dataset.

image

Run ./create_db.sh to index all images for searching. It will take some time. Be patient.

And then you are done. ./run.sh path_to_test_image_or_url 1 will give you top20 similar faces ordered by score (less is better)

pavelgonchar avatar Sep 16 '16 15:09 pavelgonchar

@pavelgonchar hello How much time does it cost on 250K dataset to do one search? Can you describe the performance infomation. Thx

aceimnorstuvwxz avatar Nov 01 '16 14:11 aceimnorstuvwxz

@rustch3n the search takes 1-2 secs on a good GPU. Approximate nearest neighbor or k-nearest neighbor makes it really fast on CPU with a bit loss of quality. I can update my code if you are interested in running this on CPU only

pavelgonchar avatar Nov 08 '16 20:11 pavelgonchar

hi pavel,

How much time taken to complete ./create_db.sh file. we are able to run this file with out any error but we  are  not getting any thing since 2 hrs.

13233 images found <vggface.Model instance at 0x7f0374868f38> WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/tensorflow/python/util/tf_should_use.py:118: initialize_all_variables (from tensorflow.python.ops.variables) is deprecated and will be removed after 2017-03-02. Instructions for updating: Use tf.global_variables_initializer instead. WARNING:tensorflow:From /usr/local/lib/python2.7/dist-packages/tensorflow/python/util/tf_should_use.py:118: initialize_local_variables (from tensorflow.python.ops.variables) is deprecated and will be removed after 2017-03-02. Instructions for updating: Use tf.local_variables_initializer instead. 2018-02-15 11:46:13.656483: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA

Sumanth214 avatar Feb 15 '18 08:02 Sumanth214