In [1]:

    

cd ..


    

    




/afs/inf.ed.ac.uk/user/s08/s0805516/repos/neukrill-net-work

In [2]:

    

import neukrill_net.utils


    

In [3]:

    

settings = neukrill_net.utils.Settings('settings.json')


    

In [4]:

    

imagedict = settings.image_fnames['train']


    

In [5]:

    

classes = list(imagedict.keys())
# copy the list
temp_classes = classes[:]
while len(temp_classes) > 0:
    # take three off the temp list on each iteration and print them
    # with tabs in between, catching the last few with the if
    if len(temp_classes) < 3:
        print("\t".join([temp_classes.pop() for i in range(len(temp_classes))]))
    else:
        print("\t".join([temp_classes.pop() for i in range(3)]))


    

    




protist_noctiluca	polychaete	tunicate_doliolid
hydromedusae_shapeA_sideview_small	fish_larvae_myctophids	copepod_calanoid_large
hydromedusae_other	protist_dark_center	acantharia_protist_big_center
hydromedusae_solmaris	fish_larvae_thin_body	tunicate_partial
jellies_tentacles	hydromedusae_shapeA	hydromedusae_shapeB
diatom_chain_tube	shrimp_caridean	copepod_cyclopoid_copilia
siphonophore_other_parts	diatom_chain_string	echinoderm_larva_pluteus_brittlestar
ctenophore_cydippid_no_tentacles	euphausiids_young	tunicate_doliolid_nurse
chaetognath_other	fish_larvae_medium_body	hydromedusae_bell_and_tentacles
fecal_pellet	tunicate_salp	ephyra
hydromedusae_narcomedusae	shrimp_sergestidae	copepod_cyclopoid_oithona_eggs
protist_fuzzy_olive	siphonophore_calycophoran_rocketship_young	invertebrate_larvae_other_B
decapods	unknown_blobs_and_smudges	radiolarian_chain
euphausiids	hydromedusae_typeD_bell_and_tentacles	echinoderm_larva_pluteus_urchin
detritus_other	invertebrate_larvae_other_A	copepod_calanoid_frillyAntennae
protist_other	unknown_sticks	echinoderm_larva_pluteus_typeC
tunicate_salp_chains	fish_larvae_leptocephali	pteropod_butterfly
siphonophore_calycophoran_sphaeronectes_stem	appendicularian_s_shape	copepod_cyclopoid_oithona
siphonophore_physonect	appendicularian_slight_curve	stomatopod
radiolarian_colony	hydromedusae_aglaura	heteropod
trochophore_larvae	hydromedusae_solmundella	chordate_type1
siphonophore_calycophoran_rocketship_adult	copepod_calanoid_flatheads	copepod_calanoid_eucalanus
hydromedusae_liriope	trichodesmium_puff	echinopluteus
fish_larvae_deep_body	siphonophore_calycophoran_abylidae	copepod_calanoid_eggs
copepod_calanoid_large_side_antennatucked	copepod_other	acantharia_protist_halo
hydromedusae_h15	copepod_calanoid	siphonophore_physonect_young
detritus_filamentous	ctenophore_cydippid_tentacles	artifacts_edge
shrimp_zoea	unknown_unclassified	hydromedusae_narco_dark
copepod_calanoid_small_longantennae	chaetognath_sagitta	echinoderm_larva_pluteus_early
hydromedusae_typeF	trichodesmium_bowtie	hydromedusae_haliscera_small_sideview
echinoderm_larva_seastar_brachiolaria	protist_star	hydromedusae_narco_young
hydromedusae_typeD	hydromedusae_typeE	acantharia_protist
artifacts	copepod_calanoid_octomoms	fish_larvae_very_thin_body
tornaria_acorn_worm_larvae	siphonophore_partial	amphipods
shrimp-like_other	chaetognath_non_sagitta	appendicularian_straight
hydromedusae_haliscera	detritus_blob	appendicularian_fritillaridae
hydromedusae_sideview_big	crustacean_other	pteropod_theco_dev_seq
ctenophore_cestid	pteropod_triangle	ctenophore_lobate
trichodesmium_tuft	echinoderm_seacucumber_auricularia_larva	hydromedusae_partial_dark
siphonophore_calycophoran_sphaeronectes	siphonophore_calycophoran_sphaeronectes_young	echinoderm_larva_seastar_bipinnaria
trichodesmium_multiple

In [6]:

    

import skimage.io
import numpy as np


    

In [7]:

    

image_array_dict = {}
for i,c in enumerate(classes):
    #print("class: {0} of 120: {1}".format(i,c))
    image_fpaths = imagedict[c]
    # dictionary will contain lists of numpy arrays
    image_array_dict[c] = []
    # iterate over all paths adding to list in dictionary
    for fpath in image_fpaths:
        image_array_dict[c].append(skimage.io.imread(fpath))


    

In [8]:

    

dimensions = [image_array.shape for imagelist in list(image_array_dict.values()) for image_array in imagelist]


    

In [9]:

    

%matplotlib inline
import matplotlib.pyplot as plt


    

In [10]:

    

plt.rcParams['figure.figsize'] = 12, 8
plt.rcParams['axes.grid'] = True
plt.set_cmap('brg')


    

    






<matplotlib.figure.Figure at 0x7ff376c96890>

In [11]:

    

x,y = zip(*dimensions)
plt.title("Histogram of Square Root of Area")
plt.grid()
plt.ylabel("Frequency")
plt.xlabel("Square root of area")
h = plt.hist(x, bins=50)
plt.savefig("plots/hist_sqrt_area.svg")
plt.savefig("plots/hist_sqrt_area.png")


    

In [12]:

    

plt.clf()
plt.scatter(x,y,s=0.1,c='black',alpha=0.8)
plt.xlabel("Size of x")
plt.ylabel("Size of y")
plt.title("Scatter Plot of Image Dimensions")
plt.savefig("plots/scatter_dimensions.svg")
plt.savefig("plots/scatter_dimensions.png")


    

In [13]:

    

# build a list of dimensions
dimensions = [neukrill_net.image_processing.landscapise_image(image_array).shape
                  for imagelist in list(image_array_dict.values())
                  for image_array in imagelist]


    

In [14]:

    

x,y = zip(*dimensions)


    

In [16]:

    

plt.clf()
plt.scatter(x,y,s=0.1,alpha=0.8)
plt.xlabel("Size of x")
plt.ylabel("Size of y")
plt.title("Scatter Plot of Re-Oriented Image Dimensions")
plt.savefig("plots/landscape_scatter_dimensions.svg")
plt.savefig("plots/landscape_scatter_dimensions.png")


    

In [17]:

    

# dictionary of classes as keys and number of examples for values
n_examples = {}
for c in classes:
    n_examples[c] = len(imagedict[c])


    

In [18]:

    

plt.title("Histogram of Class Sizes")
plt.ylabel("Frequency")
plt.xlabel("Number of Examples")
h=plt.hist(list(n_examples.values()),bins=50)
plt.savefig("plots/class_hist.svg")
plt.savefig("plots/class_hist.png")


    

In [ ]:

    

ks = [k for k in classes if n_examples[k] < 50]
lessthan50 = [n_examples[k] for k in ks]
ks,lessthan50 = zip(*sorted(zip(ks,lessthan50),key=lambda x: x[1],reverse=True))
fig, ax = plt.subplots()
ind = range(len(lessthan50))
ax.bar(ind,lessthan50)
ax.set_xticks(np.array(ind)+0.5)
ax.set_xticklabels(ks,rotation=90)
ax.set_title("Classes with less than 50 examples.")
plt.ylabel("Number of examples")
plt.savefig("plots/classes_below50.svg")
plt.savefig("plots/classes_below50.png")
plt.show()


    

In [ ]: