The Computer Vision Bootcamp

1 Introduction
2 What We’re Building
3 Vision Transformers vs Convolutional Neural Networks
4 Quadratic Operations
5 Introduction to ViTs and Joint Training with Embeddings
6 Understanding Attention Mechanisms, Brief Summary
7 Understanding the Full ViT Pipeline
8 Introduction to Prompt Encoders for SAM
9 SAM AutoPrompt Mode
10 SAM Manual Click Mode
11 ViT Embeddings inside SAM
12 Calculating Attention Score for Vision Transformers in SAM
13 How SAM is Trained
14 Calculating Prompt Self Attention for SAM
15 Prompt Image Cross Attention
16 Image to Prompt Cross Attention
17 (Optional) Finishing SAM Example Part 1
18 (Optional) Finishing SAM Example Part 2
19 Creating our SagemakerAI Domain
20 Starting Domain and Understanding Pricing
21 Installing Libraries
22 Stopping Instances and Servers
23 Downloading the SAM Model from Meta
24 Updating IAM Permissions
25 Importing Libraries
26 Understanding how we use Rekognition with SAM
27 Defining Helper Functions
28 Clarification Regarding Helper Functions
29 Rekognition Detection and Filtering
30 Initialise SAM Model from S3
31 Main Processing Function Part 1
32 Main Processing Function Part 2
33 Running the Main Processing Cell
34 Visualizing Rekognition Detections
35 Visualize All SAM Masks
36 Visualizing Match Quality IOU Scores Part 1
37 Visualizing Match Quality IOU Scores Part 2
38 Visualizing Image Segmentations with Bounding Boxes
39 Visualizing Masks and Labels Without Bounding Boxes
40 Visualizing Segementations in Black and White Masks
41 Exercise: Imposter Syndrome
42 Saving Metadata to S3
43 Save Images to S3
44 Saving Individual Masks to S3
45 Adding a GPU Server to our Notebook and AWS Quotas
46 Testing Our Full Pipeline
47 Minor Corrections
48 Productionizing + Cleanup
49 Thank You!