kahst/birdclef2017

Source code of the TUCMI submission to BirdCLEF2017

birdclef2017's Issues

How to obtain the dataset?

On the official website of imageclef there is no information on how to obtain the BirdCLEF2017 dataset. Where does one can obtain it from in the right format to use with your code?

PKL document could not be found on the server

Hi! When I am trying to run the code, there seems to be a running error:
Traceback (most recent call last):
File "birdCLEF_test.py", line 237, in
loadParams(-1, filename=PRETRAINED_MODEL)
File "birdCLEF_test.py", line 232, in loadParams
params = pickle.load(f)
File "C:\Program Files\Python\Python27\lib\pickle.py", line 1384, in load
return Unpickler(file).load()
File "C:\Program Files\Python\Python27\lib\pickle.py", line 864, in load
dispatchkey
KeyError: '<'

Can you please advise how can I download this file: birdCLEF_TUCMI_Run1_Model.pkl
Because the file was not found on this URL: https://box.tu-chemnitz.de/index.php/s/iPUsAA94KPtWaVf/download
I am getting the following error: The document could not be found on the server. Maybe the share was deleted or has expired?
Can you please advise new link for the pkl document?

Error in training on own data

I got the following error on trying to train with my own data. I followed the instructions in README.md .
Can you explain what is happening?

Traceback (most recent call last):
File "birdCLEF_train.py", line 791, in
loss = train_net(image_batch, target_batch, lr)
File "/usr/local/lib/python2.7/dist-packages/theano/compile/function_module.py", line 871, in call
storage_map=getattr(self.fn, 'storage_map', None))
File "/usr/local/lib/python2.7/dist-packages/theano/gof/link.py", line 314, in raise_with_op
reraise(exc_type, exc_value, exc_trace)
File "/usr/local/lib/python2.7/dist-packages/theano/compile/function_module.py", line 859, in call
outputs = self.fn()
MemoryError:
Apply node that caused the error: Elemwise{sqr,no_inplace}(Elemwise{Sub}[(0, 0)].0)
Toposort index: 156
Inputs types: [TensorType(float64, 4D)]
Inputs shapes: [(128, 128, 64, 128)]
Inputs strides: [(8388608, 65536, 1024, 8)]
Inputs values: ['not shown']
Outputs clients: [[Sum{axis=[0, 2, 3], acc_dtype=float64}(Elemwise{sqr,no_inplace}.0)]]

About sort data

Hello kahst.
I met the problem about 'gbk' codec can't decode bytes in position 19566-19567: illegal multibyte sequence when I read the xml file .Do you get the error too.Or just I use the Chinese to do comment and cause the error ?
Thank you

Problem while training my own data

Hello,
I got the following error on trying to train with my own data.

cannot reshape array of size 0 into shape (0,newaxis)
Apply node that caused the error: Reshape{2}(Pool{ignore_border=True, mode='max', ndim=2}.0, MakeVector{dtype='int64'}.0)
Toposort index: 247
Inputs types: [TensorType(float64, 4D), TensorType(int64, vector)]
Inputs shapes: [(0L, 1024L, 4L, 8L), (2L,)]
Inputs strides: [(262144L, 256L, 64L, 8L), (8L,)]
Inputs values: [array([], shape=(0L, 1024L, 4L, 8L), dtype=float64), array([ 0, -1], dtype=int64)]
Outputs clients: [[InplaceDimShuffle{1,0}(Reshape{2}.0), Dot22(Reshape{2}.0, W)]]

evaluation doubt

About getting magspec

In code birdCLEF_spec.py Here's a bunch of code
#fix shape to 512x256 pixels without distortion magspec = magspec[:256, :512] temp = np.zeros((256, 512), dtype="float32") temp[:magspec.shape[0], :magspec.shape[1]] = magspec magspec = temp.copy() magspec = cv2.resize(magspec, (512, 256))
However it cannot fix the array's shape to 512*256, so what's happening? And why should I do this?

birdCLEF_sort_data.py creates ? in folder names

After executing birdCLEF_sort_data.py the train and val subfolders for birds contain question marks (?). Which are not existing in the labels file.

error training in own data

I got the following error on trying to train with my own data. I followed the instructions in README.md .

ERROR (theano.gof.opt): Optimization failure due to: local_abstractconv_check
ERROR (theano.gof.opt): node: AbstractConv2d{convdim=2, border_mode='half', subsample=(2, 2), filter_flip=True, imshp=(None, 1, 256, 512), kshp=(64, 1, 7, 7), filter_dilation=(1, 1), num_groups=1, unshared=False}(input, W)
ERROR (theano.gof.opt): TRACEBACK:
ERROR (theano.gof.opt): Traceback (most recent call last):

theano.gof.opt.LocalMetaOptimizerSkipAssertionError: AbstractConv2d Theano optimization failed: there is no implementation available supporting the requested options. Did you exclude both "conv_dnn" and "conv_gemm" from the optimizer? If on GPU, is cuDNN available and does the GPU support it? If on CPU, do you have a BLAS library installed Theano can link against? On the CPU we do not support float16.

Can you explain this issue.
I built my project in windows 10 x64 bit on the CPU

How can I download the sound dataset?

Hi,

could you tell me how can i download the sound data?

Thanks.

birdCLEF_evaluate.py

I have a doubt in this the maximum prediction =110 represents the no of classes?
if we reduce the maximum prediction there is slight increase in accuracy is this way is correct can you explain this ?