weklund

LMPD

UNMANNED VEHICLE,GROUND	1	Each	$298,861.00	Q	3	Oct 4, 2018 12:00:00 AM
HELMET,FLYER'S	10	Each	$886.57	Q	3	Dec 8, 2016 12:00:00 AM
SPECTACLES SET,BALLISTIC AND LASER PROTECTIVE	35	Each	$28.27	D	1	May 2, 2010 12:00:00 AM
SPECTACLES,BALLISTIC AND LASER PROTECTIVE	50	Each	$20.16	D	1	Jul 30, 2010 12:00:00 AM
SPECTACLES KIT,BALLISTIC AND LASER PROTECTIVE	45	Kit	$15.16	D	1	Oct 28, 2009 12:00:00 AM
HELMET,FLYER'S	1	Each	$449.88	C		Mar 2, 2010 12:00:00 AM
HELMET,ADVANCED COMBAT	18	Each	$276.71	D	1	Nov 23, 2009 12:00:00 AM
HELMET,ADVANCED COMBAT	9	Each	$223.69	D	1	Sep 30, 2009 12:00:00 AM

weklund

def path_to_tensor(path):
    
    img = image.load_img(path, target_size = (224,224))
    x= image.img_to_array(img)
    return np.expand_dims(x, axis=0)

def paths_to_tensor(paths):
    list_of_tensors = [path_to_tensor(path) for path in tqdm(paths)]
    return np.vstack(list_of_tensors)
  

weklund

def load_dataset(path):
    data = load_files(path, shuffle=True)
    img_files = np.array(data['filenames'])
    img_targets = np_utils.to_categorical(np.array(data['target']), 3)
    return img_files, img_targets
  
train_files, train_labels = load_dataset('data/train')
valid_files, valid_labels = load_dataset('data/valid')

def load_dataset_no_shuffle(path):

weklund

from keras.callbacks import ModelCheckpoint

checkpoint_inception = ModelCheckpoint(
    save_best_only = True, 
    verbose = 1, 
    filepath = 'saved_models/weights.best.from_inception_resnet_v2.hdf5'
)

model.fit(
    train_features_inception, 

weklund

submission_inception_resnet = pd.DataFrame({'Id':test_files, 'task_1':test_predictions_task1,'task_2':test_predictions_task2})
pd.DataFrame.to_csv(submission_inception_resnet, 'submission.csv', index=False)

preds_path = sys.argv[1]
    
thresh = 0.5

# get ground truth labels for test dataset
truth = pd.read_csv('ground_truth.csv')
y_true = truth.as_matrix(columns=["task_1", "task_2"])

weklund

model = InceptionResNetV2(weights = 'imagenet', include_top = False)
model = Sequential()

model.add(GlobalAveragePooling2D(input_shape = train_features_inception.shape[1:]))
model.add(Dropout(0.2))
model.add(Dense(1024, activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(512, activation = 'relu'))
model.add(Dropout(0.2))
model.add(Dense(128, activation = 'relu'))

weklund

Keybase proof

I hereby claim:

• I am weklund on github.
• I am weseklund (https://keybase.io/weseklund) on keybase.
• I have a public key ASAMJSGZP-tbTZy8_qO7Vbe5UN8iJk8Tm5VG9YwLTEXGDgo

To claim this, I am signing this object:

weklund

// Use Gists to store code you would like to remember later on
console.log(window); // log the "window" object to the console