EchoPhaseDetection

Code files to accompany the paper "Multibeat Echocardiographic Phase Detection Using Deep Neural Networks". For more details about this study, including access to the dataset, please visit: https://intsav.github.io/phase_detection.html

Step 1

Place the videos from your dataset in the following directories:

| code/data/test

...

| code/data/train

...

Step 2

Generate the target labels and place in the data folder with the filename 'labels.csv'.

Csv format:

Video frame name	Label
video1_1_1	0.92
video1_1_2	0.83

Step 3

Produce a csv file containing the filenames of your training and validation sets, save to the data folder with the name 'video_info.csv'.

Csv format:

Dataset	Video filename	Number of frames
test	video1_1	30
train	video1_2	30

Step 4

Run train.py with the following args: sequence_length image_height image_width batch_size number_of_epochs

For example:

$ python train.py 30 112 112 2 1000

Step 5

Generate predictions from data/predictions/run_predictions.py using the saved model with the best weights from training.

Trained model weights

Download trained model weights