Closed JGuillaumin closed 6 years ago
Hey, some ideas:
nodejs --version
: v0.10.25
npm --version
: v0.10.25
I create my own launch
and get_app
(in server.py
) that handles input_array
instead of input_folder
as argument.
To display images (above feature maps), I save images from input_array
in temp_folder
, and it works !
@app.route('/inputs')
def get_inputs():
return jsonify(['{}.jpg'.format(idx) for idx in range(len(input_array))])
@app.route('/input-file/<path>')
def get_input_file(path):
temp_path = '{}/{}'.format(temp_folder, path)
imsave(temp_path, input_array[int(path.split('.')[0])])
return send_from_directory(abspath(temp_folder), path)
I would like to add options for handling Theano and Tensorflow ordering (now, only TensorFlow) before committing on my forked repo.
I would like to add some new features ... and I think I have to work on the JavaScript part . But I have never worked with JS. I hope I will have enough time in few weeks to learn JS and to work on those new features.
Closing, no one's reported the same installation error in over a year :)
Hi,
I would like to modify some parts of quiver :
quiver_engine.server.launch()
) a numpy array that corresponds to severals samples, with shape[nb_samples, img_w, img_h, nb_channels)
. Instead of using a directory with images (.png or .jpg)So to add those options, I tried to install the development version .. but if failed when I tried to build the client.
It doesn't end .. but it displays many errors (ENOENT and EEXIST). The first ones are :
I have the sames errors when I use
sudo npm install
.Have you had errors like this ? Or I made a mistake during installation ?
Maybe a solution is to use a Docker image that provides the development version ? Someone is working with Docker ?