Skip to content

Transfer learning#75

Open
jackattack1415 wants to merge 95 commits intodevfrom
transfer_learning
Open

Transfer learning#75
jackattack1415 wants to merge 95 commits intodevfrom
transfer_learning

Conversation

@jackattack1415
Copy link
Contributor

includes all the requisite functions to run transfer learning. to use, python run_transfer_learning.py. training_parameters.yml has some of the hyperparams that can be adjusted to tune the training. saves model with highest performance in data/ref/detection_model/TRAINING_PARAMETERS['TRAINED_MODEL_NAME'].

SamBlakeman and others added 30 commits August 1, 2019 17:39
return true_positive_dict, true_labels_dict, pred_labels_dict


def evaluate_on_gpu(sess, gpu_nms_op, pred_boxes_flag, pred_scores_flag,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this seems very different than what we had at the start of the file.
Let's break the file down into multiple files with names that make clear what type of task the functions do that live inside the file.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

broken down into four utils files: tensorflow_evaluation_utils, tensorflow_image_formatting_utis, tensorflow_plotting_utils, tensorflow_training_utils

from traffic_analysis.d04_modelling.transfer_learning.tensorflow_detection_utils import read_class_names


def transfer_learn(paths, params, train_params, train_file, test_file, selected_labels):
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

very big function. Can we modularise this at all?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants