ChaiyoGCP#5 - CodeAlong Session#2 - Build Real World AI Applications with Gemini and Imagen

#ChaiyoGCP Season 5 is a Google Cloud Study Jam running from February 27 to March 29, 2025. I’m hosting two hands-on labs:

Building an AI Image Recognition App using Gemini on Vertex AI – showcasing Gemini’s multimodal capabilities by generating rich image descriptions.
Building an AI Image Generator App using Imagen on Vertex AI – leveraging the latest Imagen 3 model to create high-quality images from text prompts.

Transcript

2. Code Along session - members Titipat Achakulvisut (My) Cloud GDE and WTM Ambassador Punsiri Boonyakiat (Beat) Lead Data Engineer @ Central Food Retail Group Nhi Nguyen (Nhi) Developer Community Manager @ Google AGENDA 19:30 PM - 19:35 PM 5 mins ChaiyoGCP Season 5 Introduction 19:35 PM - 20:30 PM 55 mins Lab 1: Build an AI Image Recognition app using Gemini on Vertex AI 20:30 PM - 21:30 PM 60 mins Lab 2: Inspect Rich Documents with Gemini Multimodality and Multimodal RAG
4. Example: Tier 1 goo.gle/ChaiyoGCP
5. Example: Tier 2 goo.gle/ChaiyoGCP
6. Claim your SWAG • Submit the completion form provided in your welcome email before March 29, 2025, 11:59 PM (GMT+7), either after earning all the required badges on Google Cloud Skills Boost or setting up your public profile. • Make a post on social media (Facebook / LinkedIn / Twitter / Instagram) with your Google Cloud Skills Boost profile photo using the hashtag #ChaiyoGCP and submit the link in the completion Form. • Participants who meet the minimum badge requirements will be invited to an exclusive online quiz for a chance to win exciting swag. The quiz will take place on one of the following dates: ◦ April 3, 2025, from 8:00 PM to 8:30 PM (GMT+7) ◦ April 5, 2025, from 10:00 AM to 10:30 AM (GMT+7) • The top 700 participants with the highest scores will win swag
7. More points = A chance to get swag • Skill Badges: Earn 2 bonus points for each additional skill badge completed within each Tier. • Early Submission: First 300 participants who submit the completion form and fulfill the requirements will receive extra 10 points. • The Online Quiz: 20 questions, with each correct answer awarding 1 point. The top 700 participants to submit and pass the online quiz will win a swag package! The top 3 participants who complete the most labs will win extra swag.
8. Option 1: April 3, 2025, from 8:00 PM to 8:30 PM (GMT+7) Option 2: April 5, 2025, from 10:00 AM to 10:30 AM (GMT+7) March Follow up Deadline Kick-off & Code Along events Online Asking support via Discord Community or email to chaiyogcp@google.com Online Recording in Youtube: ‘GDG Cloud Bangkok’ channel Social Sharing Hashtag #ChaiyoGCP Welcome Email: “[#ChaiyoGCP] ยินดี ต้อนรับสู่ #ChaiyoGCP! การ ผจญภัยในการเรียนรู้ของคุณ เริ่มต้นแล้ว!” Redeem 350 credits on Cloud Skill Boosts [In-person event] ChaiyoGCP Season 5 Kick Off and Generative AI in Action Code Along sessions Deadline to fill Completion Form Swags (8-10 weeks after result email) Make your profile public Swag will be delivered exclusively to eligible participants with addresses within Thailand. 󰑆 Result email will be sent within 1 - 2 weeks March 1 April March 11 March 29 Stay Tuned!! ! !! Running lab together from selected badge Join Discord Community - goo.gle/chaiyoGCP-chat Feb 27 Start registration Offline Quiz Apr 3 or Apr 5 Complete online quiz !!!
9. Punsiri Boonyakiat (Beat) Lead Data Engineer @ Central Food Retail Group [LAB] Build an AI Image Recognition app using Gemini on Vertex AI
10. Build an AI Image Recognition app using Gemini on Vertex AI https://www.cloudskillsboost.g oogle/course_templates/1076
11. Introduction Punsiri Boonyakiat Lead Data Engineer @ Central Food Retail Group
12. Skill Badge Information
13. Skill Badge Information
14. Skill Badge Information
15. [Lab 1] Build an AI Image Recognition app using Gemini on Vertex AI
16. • Connect to Vertex AI (Google Cloud AI platform): Learn how to establish a connection to Google's AI services using the Vertex AI SDK. • Load a pre-trained generative AI model -Gemini: Discover how to use a powerful, pre-trained AI model without building one from scratch. • Send image + text questions to the AI model: Understand how to provide input for the AI to process. • Extract text-based answers from the AI: Learn to handle and interpret the text responses generated by the AI model. • Understand the basics of building AI applications: Gain insights into the core concepts of integrating AI into software projects. Build an AI Image Recognition app using Gemini on Vertex AI
17. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226
18. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Step1 : Create new Python File and Save as genai.py
19. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Image Recognition Result Run genai.py
20. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Code Explanation นําเข้าไลบรารี vertexai ในภาษา Python pip install vertexai
21. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Code Explanation เลือก Google Cloud Project และ location ที่ต้องการเริ่มต้น ใช้งาน vertexai
22. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Code Explanation เลือก model ของ Gemini ซึ่งเป็น Generative AI model ของ Google และรับมาไว้ในตัวแปรที่ชื่อว่า multimodal_model
23. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226 Code Explanation เรียกใช้ function generate_content Prompt = Image + Text
24. Build an AI Image Recognition app using Gemini on Vertex AI Lab: https://www.cloudskillsboost.google/course_templates/1076/labs/488226
26. Gemini Model List
27. [Lab 2] Build an AI Image Generator app using Imagen on Vertex AI
28. Build an AI Image Generator app using Imagen on Vertex AI
29. Build an AI Image Generator app using Imagen on Vertex AI เลือก Model Imagegen จาก Pretrain Image Generation Model
30. Build an AI Image Generator app using Imagen on Vertex AI เรียกใช้ function generate_image เพือสร้างรูปภาพ
32. Create images from a prompt in seconds Text to image generation with high quality and speed • Aspect ratios 1:1, 9:16, 16:9, 3:4, 4:3 • Higher Visual Detail, Prompt Adherence, and general preference Portrait of a couple laughing and holding hands as they walk through an amusement park at sunset. The background is filled with vibrant carnival, holding cotton candy, and a towering Ferris wheel.. A dynamic, high-energy animation of a superhero flying across a busy metropolis with a blurred background A close up image of four pairs of hands on a poker table A bright style- magazine shot of a woman putting on mascara. Imagen 3 Generation https://developers.googleblog.com/en/imagen-3-arrives-in-the-gemini-api/
33. Create images from a prompt in seconds A dynamic, high-energy animation of a superhero flying across a busy metropolis with a blurred background Imagen 3 Generation https://developers.googleblog.com/en/imagen-3-arrives-in-the-gemini-api/ a portrait of a sheepadoodle wearing cape
34. Imagen Model Version https://cloud.google.com/vertex-ai/generative-ai/docs/image/model-versioning
35. LINK: aistudio.google.com Let’s Try >
36. • Unified SDK for: ◦ Gemini Developer API ◦ Gemini API on Vertex AI • Change one (ish) line to switch between them. ◦ Or Environment Variables • Supports Gemini 1.5 & above • Language support ◦ Current: Python, Go, Java ◦ Future: JavaScript pip install google-genai from google import genai # Gemini Developer API client = genai.Client(api_key="YOUR_API_KEY") # Vertex AI API client = genai.Client( vertexai=True, project="your-project-id", location="us-central1", ) Google Gen AI SDK
37. Titipat Achakulvisut (My) Biomedical Engineering @ Mahidol University [LAB] Inspect Rich Documents with Gemini Multimodality and Multimodal RAG
38. Inspect Rich Documents with Gemini Multimodality and Multimodal RAG https://www.cloudskillsboost.g oogle/course_templates/981
39. Introduction Titipat Achakulvisut Biomedical Engineering @ Mahidol University GitHub: https://github.com/titipata Website: biodatlab.github.io
40. Inspect Rich Documents with Gemini Multimodality and Multimodal RAG • Lab 2: Using Gemini for Multimodal Retail Recommendations • Lab 3: Multimodal Retrieval Augmented Generation (RAG) using the Gemini API in Vertex AI Titipat Achakulvisut Biomedical Engineering @ Mahidol University GitHub: titipata Website: biodatlab.github.io
41. Skill Badge Information Lab 2 Lab 3 Goes to https://goo.gle/chaiyoGCP
42. [Lab 2] Use Gemini for Multimodal Retail Recommendation
43. Using multimodal functionality with Gemini 1.5 Pro Describe what’s visible in this room and the overall atmosphere: Prompt Image Generate content response
44. Using multimodal functionality with Gemini 1.5 Pro Recommend a new piece of furniture for this room: Prompt Image and explain the reason in detail Prompt Generate content response
45. Combining multiple images Consider the following chairs: Room: You’re an interior designer. For each chair explain … Stacking in a prompt
46. [Lab 3] Multimodal Retrieval Augmented Generation (RAG) using the Gemini API in Vertex AI
47. Let’s imagine if we have a report in PDF format, how can we use it to answer the given query? Extract and store metadata from a document
48. Extract and store metadata from a document Plain text Graph Table Text metadata - Text - Chunk text - Chunk number - Text embedding Image metadata - Image description (Generated with Gemini) - Text + Image embedding - Image embedding - Text embedding Plain text
49. Text Search / Image Description Search Text query Retrieve based on description response Text metadata Image metadata Prompt: Answer with the given context Questions: {} Context: {} Answer: Retrieval Augmented Generation (RAG)
50. Image Search Retrieve based on image Image query response Prompt: Answer with the given context Questions: {} Context: {} Answer:
51. Multimodal retrieval augmented generation (RAG) Retrieve similar text Text Query response Prompt: Answer with the given context Questions: {} Context: {} Answer: Retrieve similar image (description) Retrieval Augmented Generation (RAG) Text metadata Image metadata
52. Q & A
53. **เขารวมได ทุกเพศทุกวัย
54. Women Techmakers Bangkok
55. LIKE & SHARE