n_state_space=N_STATE_SPACE, base_impacts=BASE_IMPACTS, agg_weeks=4, save_calc=False)
print('-------------------------------- ' + str(device) + ' --------------------------------')
if device == 'cuda':
print(torch.cuda.get_device_name(0))
# initialize agents and network optimizers and store them in dicts
optimizers, all_agents = init_agents(ACTION_SPACE, N_STATE_SPACE, REQUIRED_NEURAL_NETS, ACT_AGT, LEARNING_RATE, device)
# save initial agents
torch.save(all_agents, (SAVE_DIR / 'agents_init'), _use_new_zipfile_serialization=False)
# initialise loop variables
total_rewards = {'FSC': [], 'Shell': []}
batch_returns = {'FSC': [], 'Shell': []}
batch_actions = create_dict(REQUIRED_NEURAL_NETS, ACT_AGT)
batch_states = {'FSC': np.empty([BATCH_SIZE, LENGTH_EPISODE, N_STATE_SPACE['FSC']]),
'Shell': np.empty([BATCH_SIZE, LENGTH_EPISODE, N_STATE_SPACE['Shell']]),
'Gov': np.empty([BATCH_SIZE, LENGTH_EPISODE, N_STATE_SPACE['Gov']])}
batch_rewards = {'FSC': np.empty([BATCH_SIZE, LENGTH_EPISODE, 1]),
'Shell': np.empty([BATCH_SIZE, LENGTH_EPISODE, 1])}
support_calc = {'Shell': np.empty([BATCH_SIZE, LENGTH_EPISODE, len(AGENTS) + 1]),
'Gov': np.empty([BATCH_SIZE, LENGTH_EPISODE, len(AGENTS) + 1])}
reward_shell_calc = np.empty([BATCH_SIZE, LENGTH_EPISODE, 4])
batch_count = 0
optim_count = 0
step_count = 0
ep = 0
times = []
# loop over all episodes (= rollouts)
while ep < NUM_EPISODES:
import functions as f
import deployed
app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = 8192 * 8192
app.config['UPLOAD_EXTENSIONS'] = ['.jpg', '.png']
app.config['UPLOAD_PATH'] = 'static/uploads'
app.config['IMAGE_PATH'] = 'static/images'
# code to download and store all images; no longer necessary after run for first time
# since we've now saved all images in the repository and pushed to github
# f.download_images(f.create_dict('static/image_links.csv')[0], "static/images/")
# code to load all necessary info for each plant disease
link_dict=f.create_link_dict("static/info_links.csv")
plant_disease_dict, diseases = f.create_dict('static/disease_info.csv')
# results=['Potato Late Blight', 'Tomato Leaf Mold', 'Tomato Late Blight', 'Peach Healthy']
@app.route('/')
def index():
return render_template('index.html')
@app.route('/display')
def display():
files = os.listdir(app.config['UPLOAD_PATH'])
results=[]
for file in files:
results.append(diseases[deployed.predict_new(os.path.join(app.config['UPLOAD_PATH'], file))])
return render_template('display.html', files=files, pd_dict=plant_disease_dict, results=results, num=len(files), link_dict=link_dict)
def test_pickle_dict(self):
functions.create_dict('test_dict')
files = listdir(path.dirname(path.abspath(__file__)))
self.assertIn('test_dict', files)