Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

AI for Social Good - Saving the Planet with Data Science

595 views

Published on

Presentation by Ganes Kesari at Strata, London on 30-Apr-2019

Published in: Data & Analytics
  • Be the first to comment

  • Be the first to like this

AI for Social Good - Saving the Planet with Data Science

  1. 1. AI for Social Good: Saving the planet with Data Science Ganes Kesari, Gramener @kesaritweets
  2. 2. 2Gramener Introduction Co-founder & Head of Analytics 100+ Clients Insights as Stories Help apply & adopt Analytics Our data science platform, Gramex is now open-sourced! “Simplify Data Science for all”
  3. 3. 3Gramener What ails our Biodiversity? Now Extinct! Movie copyright: Blue Sky Studios Image by AngieToh from Pixabay Now Extinct…
  4. 4. 4Gramener Can we save our Biodiversity? Source: Image1 ; Image2; Image3; Image4
  5. 5. 5Gramener Spotting, Identifying and Counting Animals to Save them Live Case Studies from Gramener’s partnership with Microsoft AI for Earth https://www.microsoft.com/en-us/ai/ai-for-earth Climate Agriculture Biodiversity Water
  6. 6. 6Gramener The path to protecting animal species Detect
  7. 7. 7Gramener Mugshots in the Forest Source: Giphy (https://media.giphy.com/media/1k10MvscEjChyyHtmQ/giphy.mp4)
  8. 8. 8Gramener Sample Camera trap pictures
  9. 9. 9Gramener Input Output Identify features to teach model Traditional Machine Learning Person Name 9 Deep Learning Input Output Model automatically identifies features to learn Person Name https://www.cs.toronto.edu/~ranzato/publications/taigman_cvpr14.pdf Deep Learning to the rescue
  10. 10. 10Gramener Not very different from how we learn Training ..Versatile Detection!
  11. 11. 11Gramener Detecting animals: Approach https://towardsdatascience.com/build-your-first-deep-learning-classifier-using-tensorflow-dog-breed-example-964ed0689430
  12. 12. 12Gramener Detecting animals: Approach
  13. 13. 13Gramener The path to protecting animal species Detect Identify
  14. 14. 14Gramener Where have the Salmon Gone? Source: Giphy (https://media.giphy.com/media/QM5GJO6J8lDfa/giphy.mp4)
  15. 15. 15Gramener Underwater Salmon snapshots
  16. 16. 16Gramener Approach
  17. 17. 17Gramener The Model in Action Video
  18. 18. 18Gramener The path to protecting animal species Detect Identify …from videos …from pictures
  19. 19. 19Gramener Can we Classify All species on Earth? Source: Giphy (https://media.giphy.com/media/CFk1wEH7Cke0E/giphy.mp4)
  20. 20. 20Gramener iNaturalist: Fostering Citizen Science https://play.google.com/store/apps/details?id=org.inaturalist.android&hl=en
  21. 21. 21Gramener Building the Model Inception v4
  22. 22. 22Gramener The path to protecting animal species Detect Identify Count
  23. 23. 23Gramener Let’s Save the Penguins Source: Giphy (https://giphy.com/gifs/push-bwLowbhUWm2lO)
  24. 24. 24Gramener Tracking the Penguins https://www.zooniverse.org/projects/penguintom79/penguin-watch/about/research ~100 cameras in 16 locations, and hourly images over years Crowd-sourced annotations to identify the penguins
  25. 25. 25Gramener Crowd-sourced Annotations https://www.zooniverse.org/projects/penguintom79/penguin-watch/classify
  26. 26. 26Gramener Counting Crowds – Bounding boxes vs Density estimates Occlusion Density Difference Perspective Distortion Camera angle Preserve spatial information Localize count Handle scale variations No longer looking for a head!
  27. 27. 27Gramener Approach https://arxiv.org/pdf/1707.09605.pdf Input an image Density Map Estimate the countSplit into 9 patches
  28. 28. 28Gramener Approach • High-level prior to classify image into buckets • Density estimation to create the density map • NC6 v3 virtual machine with V100 GPU card • Trained for 200 epochs, MAE for the model : ~10.5 https://arxiv.org/pdf/1707.09605.pdf Work done by Gramener in partnership with Microsoft AI for Earth
  29. 29. 29Gramener Results https://arxiv.org/pdf/1707.09605.pdf
  30. 30. 30Gramener The path to protecting animal species Detect Identify Count Protect
  31. 31. 31Gramener Saving elephants in the Wild Source: Giphy (https://media.giphy.com/media/1AHZzdXVTYDtJVTO5a/giphy.mp4)
  32. 32. 32Gramener Can you spot the elephants? https://www.savetheelephants.org/project/tsavo-aerial-defence/
  33. 33. 33Gramener Approach Region Proposal Network Region of Interest Pooling Regional CNN FASTER RCNN
  34. 34. 34Gramener Model Results https://www.savetheelephants.org/project/tsavo-aerial-defence/
  35. 35. 35Gramener Join the Revolution! AnnotateShare your pictures Spread the word
  36. 36. 36Gramener 6 Deep Learning Takeaways: When the Rubber hits the Road Scout for the right Data Watch out for labelling Educate on Accuracy Augment with human inputs Build into a workflow Budget refresh & upkeep
  37. 37. Session deck with model details, Github repos & bonus case study at @kesaritweetsgramener.com @kesari gkesari.com/strata
  38. 38. Rate today’s session Session page on conference website O’Reilly Events App

×