Model_maker_training#

Model_maker_training.load_data(csv_file_path, images_dir)#

Loads data from the csv file.

Parameters:
  • csv_file_path – The file path to the csv file.

  • images_dir – The directory that contains images.

Returns:

A Dataloader containing the data of training, validation and test set.

Model_maker_training.train_model(train_data, validation_data)#

Fine tunes the EfficientDet-Lite0 model on the given dataset.

Parameters:

train_data – Training dataset, in the form of Dataloader. validation_data: Validation dataset, in the form of Dataloader.

Returns:

A trained model.

Model_maker_training.preprocess_image(image_path, input_size)#

Preprocess the input image to feed to the TFLite model

Parameters:
  • image_path – path to the input image

  • input_size – the input size of the model

Returns:

preprocessed image in the format of (1, input_size, input_size, 3) original image size

Model_maker_training.detect_objects(interpreter, image, threshold)#

Returns a list of detection results, each a dictionary of object info.

Parameters:
  • interpreter – tflite.Interpreter

  • image – A [1, height, width, 3] Tensor of type tf.uint8.threshold: a

  • number (floating point) –

Returns:

a list of dicts

Model_maker_training.run_odt_and_draw_results(image_path, interpreter, threshold=0.5)#

Run object detection on the input image and draw the detection results

Parameters:
  • image_path – path to the input image

  • interpreter – tflite.Interpreter

  • threshold – a floating point number