Semantic Segmentation models, evaluation and advanced techniques

Navigate through the different lesson sections by using the menu on the lefthand side or the table of contents below.

Table of contents

Session Agenda

Date: 09/08/2023
Instructor(s): Lilly Thomas & Ryan Avery

Time	Topics	Presenter
08:30 - 09:00	Breakfast & Coffee / Meet and Greet	-
09:00 - 10:30	Building, compiling, and training a deep learning segmentation model to process datasets Format: Lecture and Interactive	Moderator: Vanesa Martin (NASA) Presenter: Lilly Thomas (Development Seed) Ryan Avery (Development Seed)
10:30 - 11:00	Coffee Break	-
11:00 - 13:00	Advanced Techniques for Deep Learning, metrics evaluation, data augmentation, etc Format: Lecture and Interactive	Moderator: Vanesa Martin (NASA) Presenter: Lilly Thomas (Development Seed) Ryan Avery (Development Seed)
13:00 - 13:30	Lunch	-
13:30 - 15:00	Testing different architecture and transfer learning with Convolutional Neural Networks Format: Lecture	Moderator: Vanesa Martin (NASA) Presenter: Lilly Thomas (Development Seed) Ryan Avery (Development Seed)
15:00 - 15:30	Break	-
15:30 - 17:00	TensorFlow advanced applications with Deep Learning: Object detection Regression Time Series Analysis Hyper-parameter tuning Format: Lecture	Moderator: Vanesa Martin (NASA) Presenter: Lilly Thomas (Development Seed) Ryan Avery (Development Seed)
17:00 - 17:20	Day 2 wrap up	Moderator: Vanesa Martin (NASA) Milagros Becerra (ACCA)

Content

Building, compiling, and training a deep learning segmentation model to process datasets:

• Processing for segmentation: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson5a_prep_data_ML_segmentation.html
• Segmentation with conv net: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson5b_deeplearning_segmentation_UNet.html
• Segmentation with ViT: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson5c_segmentation_ViT.html

Advanced Techniques for Deep Learning, metrics evaluation, data augmentation, etc:

• Evaluating Semantic Segmentation Models: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson6a_evaluation.html
• Dealing with limited data for semantic segmentation: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson6b_dealing_with_limited_data.html

Testing different architecture and transfer learning with Convolutional Neural Networks:

• Comparing deep learning architectures for different computer vision tasks on satellite imagery: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson7a_comparing_architectures.html
• Comparing RNNs and Transformers for Imagery: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson7b_comparing_RNN_transformer_architectures.html
• Transfer learning, fine-tuning: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson7c_transfer_learning_hyperparam_opt.html

TensorFlow advanced applications with Deep Learning:

• TensorFlow advanced applications with deep learning object detection, time-series analysis: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson8_advanced_applications.html
• Regression: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson4a_GEE_PythonAPI_TensorFlow_Regression.html
• Hyperparameter tuning: https://developmentseed.org/tensorflow-eo-training-2/docs/Lesson7c_transfer_learning_hyperparam_opt.html