def __init__(self, master=None):
self.params = params()
if not self.params.dirs:
print('config file - failed to load')
return
tk.Frame.__init__(self, master, background="yellow")
self.pack(fill="both", expand=True)
self.createWidgets()
if not self.params.files['model_trash'] == 'None':
self.cnn_trash = classifications.cnn_classification(
self.params.files['model_trash'],
im_height=self.params.neural['im_height'],
im_width=self.params.neural['im_width'])
print('load model ' + self.params.files['model_trash'])
self.cnn_taxon = classifications.cnn_classification(
self.params.files['model_taxon'],
im_height=self.params.neural['im_height'],
im_width=self.params.neural['im_width'])
print('load model ' + self.params.files['model_taxon'])
if self.params.processing['auto_start'] == 'True':
self.start()
else:
self.running_state = 'stopped'
def __init__(self, params):
self.params = params
self.cur_progress = 0
self.elapsed = 0
self.correct_RGBShift = self.params.processing['ch_shift'] == 'True'
if not self.params.files['model_trash'] == 'None':
self.cnn_trash = classifications.cnn_classification(
self.params.files['model_trash'],
im_height=self.params.neural['im_height'],
im_width=self.params.neural['im_width'])
print('load model ' + self.params.files['model_trash'])
self.cnn_taxon = classifications.cnn_classification(
self.params.files['model_taxon'],
im_height=self.params.neural['im_height'],
im_width=self.params.neural['im_width'])
print('load model ' + self.params.files['model_taxon'])
def __init__(self, params):
self.params = params
self.cur_progress = 0
self.elapsed = 0
self.correct_RGBShift = self.params.processing['ch_shift'] == 'True'
self.save_cropped = self.params.processing['save_cropped'] == 'True'
self.cnn_taxon = classifications.cnn_classification(
self.params.files['model_taxon'],
model_output_layer=self.params.neural['model_output_layer'])
self.classes = []
print('load model ' + self.params.files['model_taxon'])
def __init__(self, master=None):
self.params = params()
tk.Frame.__init__(self, master, background="green")
self.pack(fill="both", expand=True)
self.createWidgets()
self.cnn = classifications.cnn_classification(
self.params.files['model'])
print('load model ' + self.params.files['model'])
# check result folder
check_folder(folder=os.path.join(self.params.dirs['root'],
self.params.dirs['classification']),
create=True)
check_folder(folder=os.path.join(self.params.dirs['root'],
self.params.dirs['classification'],
self.params.dirs['results']),
create=True)
print('result folders are checked and created')
default=None)
parser.add_argument('-s',
action='store',
dest='s',
type=bool,
required=False,
default=False)
parser.add_argument('-o',
action='store',
dest='o',
type=str,
required=False,
default='')
# create classfier
cnn = classifications.cnn_classification()
# Parse the arguments
inargs = parser.parse_args()
path_str = os.path.abspath(inargs.i)
if inargs.b is None:
logging.info('Single image process')
cell_classifier(path_str,
cnn=cnn,
save_diag=inargs.s,
out_dir=inargs.o)
else:
logging.info('Batch execution')
batch_cell_classifier(path_str,
cnn=cnn,
cfg = train_params(data_dir, training_id=training_id)
typedict_file = os.path.join(cfg.train_dir, 'type_dict.csv')
model_file = os.path.join(cfg.train_dir, 'cnn_model.dnn')
type_dict = {}
reader = csv.DictReader(open(typedict_file, 'rt'), delimiter=':')
for row in reader:
type_dict[row['label']] = row['type']
sorted_classes = OrderedDict(sorted(type_dict.items(),
key=lambda x: x[1])).values()
# LOAD MODEL
cnn_class = classifications.cnn_classification(model_file)
num_classes = cnn_class.pred.output.shape[0]
df_test_image = pd.read_csv(cfg.test_image_list_file, delimiter=';')
df_test_text = pd.read_csv(cfg.test_text_list_file, delimiter=';')
samples = {}
contingency_table = np.zeros((num_classes, num_classes))
misclassified = []
df_images_processed = df_test_image.reset_index(drop=True).copy()
df_images_processed['predicted_type'] = None
df_images_processed['prob_taxon'] = None
pd.options.mode.chained_assignment = None # default='warn'