import matplotlib.pyplot as plt
if __name__ == "__main__":
slice_samples = 24
# Ruta y nombre del modelo guardado
ia_file_path = "/home/augusto/Desktop/GIBIO/Algoritmos/ia/trained_models/model_1_beta_large"
ia_filename = "qrs_det_model_1_epoch_10.pt"
# Archivo a testear
rec_path = "/home/augusto/Desktop/GIBIO/processed_dbs/only_MLII_agosto/mitdb"
rec_filename = "217.bin"
# Cargo modelo
model = qrs_detector.qrs_det_1_beta(24)
model.load_state_dict(torch.load(os.path.join(ia_file_path, ia_filename)))
model.eval()
# Cargo archivo a testear
data = pickle.load(open(os.path.join(rec_path, rec_filename), 'rb'))
counter = 0
while True:
this_data = torch.Tensor(
data['data'][counter * slice_samples:counter * slice_samples +
slice_samples])
this_data = this_data.view(1, 1, 24)
y_pred = model(this_data)
if realistic_check:
save_filename += "_realistic.bin"
else:
save_filename += "_norealistic.bin"
if not os.path.isfile(os.path.join(save_path, save_filename)):
# Archivos a testear
test_file_handler = open(os.path.join(test_path, test_filename), 'r')
test_files = test_file_handler.read()
test_files = test_files.split('\n')
test_files = list(filter(None, test_files))
test_filepaths = [os.path.join(test_path, s) for s in test_files]
test_file_handler.close()
# Modelo de IA a testear
model = qrs_detector.qrs_det_1_beta(slice_samples)
model.load_state_dict(
torch.load(os.path.join(ia_file_path, ia_filename)))
model.eval()
# Matriz de confusion
results = {}
results['confusion_matrix'] = []
results['specifiers'] = []
results['file_name'] = []
results['file_path'] = []
print("Entering test loop...")
if realistic_check:
print("Mode: Realistic check.")
# Donde guardar el modelo
model_path = "/home/augusto/Desktop/GIBIO/Algoritmos/ia/trained_models/model_rata_v3"
model_filename = "qrs_det_model_rata_epoch_{}.pt"
# Donde guardar historia del entrenamiento
train_progress_path = "/home/augusto/Desktop/GIBIO/Algoritmos/ia/trained_models/model_rata_v3"
train_progress_filename = "qrs_det_model_rata_training.bin"
# Si num_workers no es cero, la RAM aumenta constantemente hasta que explota el SO
train_dataloader = DataLoader(train_gen,
batch_size=batch_size,
shuffle=True,
num_workers=0)
# Definicion del modelo
model = qrs_detector.qrs_det_1_beta(train_gen.shape)
loss_fn = torch.nn.BCELoss()
learning_rate = 250e-6
optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)
print("Entering training loop...")
print("Total epochs: {}".format(total_epochs))
if train_time_in_secs == 0 and batchs_per_epoch == 0:
print("Whole training set mode.")
elif batchs_per_epoch > 0:
print("Batchs per epoch: {}".format(batchs_per_epoch))
else: