Plant AI: Project Showcase

Transcript

3. Motivation • Read a disturbing headline
4. Motivation • Read a disturbing headline Source: Times Of India
5. Motivation - Times Of India Most prominent Indian newspaper
6. Motivation • Read a disturbing headline • Researched a bit about the problem
7. Motivation • Read a disturbing headline • Researched a bit about the problem • Crop losses Source: National Academy of Science, Singh et al.
8. Motivation • Read a disturbing headline • Researched a bit about the problem • Crop losses • We 🧡 tech and doing social good
9. The Problem • Crop losses: diseases and deficiencies
11. The Project • Turns out almost all farmers have access to smartphones - Research Study by Hughes et al.
12. The Project • Turns out almost all farmers have access to smartphones • What does this do? ◦ Uses AI to diagnose diseases from plant images early
13. The Project • Turns out almost all farmers have access to smartphones • What does this do? ◦ Uses AI to diagnose diseases from plant images early ◦ Provides actionable ways to solve diseases
14. The Project • Turns out almost all farmers have access to smartphones • What does this do? ◦ Uses AI to diagnose diseases from plant images early ◦ Provides actionable ways to solve diseases ◦ Specific ways to solve
15. The Project • Turns out almost all farmers have access to smartphones • What does this do? ◦ Uses AI to diagnose diseases from plant images early ◦ Provides actionable ways to solve diseases ◦ Specific and actionable ways to solve, AI also identifies species ◦ Works Offline, PWA
16. The Project
17. The Project
18. The Project • Turns out almost all farmers have access to smartphones • What does this do? ◦ Uses AI to diagnose diseases from plant images early ◦ Provides actionable ways to solve diseases ◦ Specific and actionable ways to solve, AI also identifies species ◦ Works Offline • Approximated decrease in plant losses: 67% -> 25%
19. Existing Solutions • CropNet (Google), only Cassava plants
20. Existing Solutions • CropNet (Google): only Cassava plants • Plant Village (Hughess et al.): only a dataset
21. Existing Solutions • CropNet (Google): only Cassava plants • Plant Village (Hughess et al): only a dataset • Plant Disease detector (Ramesh et al): unsatisfactory performance
22. Existing Solutions • CropNet (Google): only Cassava plants • Plant Village (Hughess et al): only a dataset • Plant Disease detector (Ramesh et al): unsatisfactory performance • Some other disease detectors: not optimized or usable on small devices
23. How we built this? - The Model • Collected data
24. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters
25. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters
26. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters • Optimize the model to run on-device (in a PWA)
27. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters • Optimize the model to run on-device (in a PWA) • Optimized the model to be just 12 MBs
28. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters • Optimize the model to run on-device (in a PWA) • Optimized the model to be just 12 MBs
29. How we built this? - The Model • Collected data • Experimented with multiple architectures and hyperparameters • Optimize the model to run on-device (in a PWA) • Optimized the model to be just 12 MBs Wait, this is fast!💡
30. How we built this? - Web App • Offline Support PWA
31. How we built this? - Web App • Offline Support PWA • Run the optimized model with TFJS
32. How we built this? - Web App • Offline Support PWA • Run the optimized model with TFJS ◦ Individually fetch data flow graph and weights ◦ Normalize images ◦ Resize with nearest neighbour interpolation
33. How we built this? - Web App • Offline Support PWA • Run the optimized model with TFJS ◦ Individually fetch data flow graph and weights ◦ Normalize images ◦ Resize with nearest neighbour interpolation • Finally, deploy the web app to test out with real life scenarios🚀
34. Business Plan • Farmers don't get charged
35. Business Plan • Farmers don't get charged • Potential plant product companies place ads
36. Business Plan • Farmers don't get charged • Potential plant product companies place ads • After a quality check their ads appear under “How to Solve”
37. Business Plan • Farmers don't get charged • Potential plant product companies place ads • After a quality check their ads appear under “How to Solve” • Data, data and data
38. Business Plan • Farmers don't get charged • Potential plant product companies place ads • After a quality check their ads appear under “How to Solve” • Data, data and data • Collect non sensitive but useful information
39. Business Plan • Farmers don't get charged • Potential plant product companies place ads • After a quality check their ads appear under “How to Solve” • Data, data and data • Collect non sensitive but useful information • Plant images are used anonymously to improve the ML
40. Business Plan • Farmers don't get charged • Potential plant product companies place ads • After a quality check their ads appear under “How to Solve” • Data, data and data • Collect non sensitive but useful information • Plant images are used anonymously to improve the ML • Usage data is shared to plant product companies
41. Impact • Can reduce crop losses by 67% -> 25%🚀 • Tested on real scenarios • Featured on Microsoft Blog and YouTube🤗
42. Thank you! And we 🧡 open-source https://git.io/JujAg View on GitHub Research supported with Cloud TPUs from Google's TPU Research Cloud (TRC)