Slide 1

Slide 1 text

No content

Slide 2

Slide 2 text

No content

Slide 3

Slide 3 text

Motivation ● Read a disturbing headline

Slide 4

Slide 4 text

Motivation ● Read a disturbing headline Source: Times Of India

Slide 5

Slide 5 text

Motivation - Times Of India Most prominent Indian newspaper

Slide 6

Slide 6 text

Motivation ● Read a disturbing headline ● Researched a bit about the problem

Slide 7

Slide 7 text

Motivation ● Read a disturbing headline ● Researched a bit about the problem ● Crop losses Source: National Academy of Science, Singh et al.

Slide 8

Slide 8 text

Motivation ● Read a disturbing headline ● Researched a bit about the problem ● Crop losses ● We 🧡 tech and doing social good

Slide 9

Slide 9 text

The Problem ● Crop losses: diseases and deficiencies

Slide 10

Slide 10 text

No content

Slide 11

Slide 11 text

The Project ● Turns out almost all farmers have access to smartphones - Research Study by Hughes et al.

Slide 12

Slide 12 text

The Project ● Turns out almost all farmers have access to smartphones ● What does this do? ○ Uses AI to diagnose diseases from plant images early

Slide 13

Slide 13 text

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

Slide 14

Slide 14 text

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

Slide 15

Slide 15 text

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

Slide 16

Slide 16 text

The Project

Slide 17

Slide 17 text

The Project

Slide 18

Slide 18 text

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%

Slide 19

Slide 19 text

Existing Solutions ● CropNet (Google), only Cassava plants

Slide 20

Slide 20 text

Existing Solutions ● CropNet (Google): only Cassava plants ● Plant Village (Hughess et al.): only a dataset

Slide 21

Slide 21 text

Existing Solutions ● CropNet (Google): only Cassava plants ● Plant Village (Hughess et al): only a dataset ● Plant Disease detector (Ramesh et al): unsatisfactory performance

Slide 22

Slide 22 text

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

Slide 23

Slide 23 text

How we built this? - The Model ● Collected data

Slide 24

Slide 24 text

How we built this? - The Model ● Collected data ● Experimented with multiple architectures and hyperparameters

Slide 25

Slide 25 text

How we built this? - The Model ● Collected data ● Experimented with multiple architectures and hyperparameters

Slide 26

Slide 26 text

How we built this? - The Model ● Collected data ● Experimented with multiple architectures and hyperparameters ● Optimize the model to run on-device (in a PWA)

Slide 27

Slide 27 text

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

Slide 28

Slide 28 text

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

Slide 29

Slide 29 text

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!💡

Slide 30

Slide 30 text

How we built this? - Web App ● Offline Support PWA

Slide 31

Slide 31 text

How we built this? - Web App ● Offline Support PWA ● Run the optimized model with TFJS

Slide 32

Slide 32 text

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

Slide 33

Slide 33 text

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🚀

Slide 34

Slide 34 text

Business Plan ● Farmers don't get charged

Slide 35

Slide 35 text

Business Plan ● Farmers don't get charged ● Potential plant product companies place ads

Slide 36

Slide 36 text

Business Plan ● Farmers don't get charged ● Potential plant product companies place ads ● After a quality check their ads appear under “How to Solve”

Slide 37

Slide 37 text

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

Slide 38

Slide 38 text

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

Slide 39

Slide 39 text

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

Slide 40

Slide 40 text

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

Slide 41

Slide 41 text

Impact ● Can reduce crop losses by 67% -> 25%🚀 ● Tested on real scenarios ● Featured on Microsoft Blog and YouTube🤗

Slide 42

Slide 42 text

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)