Introduction and course content
Image segmentation in computer vision
Instance segmentation using Mask RCNN
Software setup
-
Brief intro to Tensorflow 2 object detection API (1:45)
-
Linux installation : How to install tensorflow 2 with GPU support (part 1) (9:09)
-
Linux installation : How to install tensorflow 2 with GPU support (part 2) (11:14)
-
Linux installation : How to install tensorflow 2 object detection API (7:07)
-
Windows installation : Installing miniconda (2:15)
-
Windows installation : Create virtual environment (1:58)
-
Windows installation : Installing tensorflow 2 object detection API (13:23)
-
Windows installation : Installing tensorflow with GPU support (23:01)
Custom data preparation
-
Choosing a dataset (0:36)
-
Linux - Exploring the dataset - Part 1 (4:24)
-
Linux - Exploring the dataset - Part 2 (5:33)
-
Windows - Exploring the dataset (5:14)
-
Annotating a custom dataset (17:04)
-
From multiple annotation files to one annotation file (7:14)
-
Transforming our dataset to tfrecord format (7:09)
Train Mask RCNN model on your local machine
-
Training on premise VS training on the cloud (1:42)
-
Downloading Mask RCNN pretrained model (3:15)
-
Finding the right configuration file (1:31)
-
Exploring the configuration file (4:23)
-
Modifying the configuration file - Part 1 (12:35)
-
Modifying the configuration file - Part 2 (3:23)
-
Running the training locally (8:34)
Train and evaluate Mask RCNN model using google AI platform
-
What is cloud computing and what is AI-Platform? (optional) (5:48)
-
Creating a Google Cloud account (5:28)
-
Downloading Google Cloud SDK (5:50)
-
Setting up a new project on google cloud platform (2:43)
-
Creating a google bucket and uploading data to it (6:20)
-
Preparing our config file for training on google cloud (5:42)
-
Checking connection to google cloud from within our local machine (3:13)
-
Exploring the training command (5:36)
-
Running the training for Mask RCNN model (3:35)
-
Checking the progress of the training job on google ai platform (3:24)
-
Running the evaluation for Mask RCNN model during the training (6:32)
-
Analyzing the results after the training of Mask RCNN model is finished (8:24)
-
Analyzing the results of the second training (8:34)
-
Further explanation of when to run your evaluation jobs (2:14)
-
Downloading the trained model and exporting the SavedModel from checkpoints (8:21)
-
Running the exported model on new examples locally (22:06)
Summary
