Home

Welcome to the orcadata wiki!

This is a place to share and collaborate, especially regarding bioacoustic analysis of real-time and archived audio data related to the Orcasound open source project. Here you can learn more about Orcasound: machine learning resources related to orcas (training sets | test sets) and access to Orcasound data -- both archived training and testing data, and real-time audio streams. You may also be interested in the synopses of projects that leverage these open data at ai4orcas.net.

Most-recent progress (within the last year or so)

2021

• Sep: J pod returns after 5 month hiatus; OrcaHello team plans for annual Microsoft hackathon (Oct 12-14); Alex Barnhill of ORCA-SPOT team joins Orcasound Slack.
• Aug: L54s heard twice on Orcasound Lab hydrophones; GSoC student projects completed, including embedding visualizations for OrcaAL, and Github action workflow for increased data flow and algorithm deployment.
• July: K pod visits (but isn't heard on Orcasound hydrophones); HALLO project develops v0.3 of KW call model, onboards Sadman Sakib
• June: Orcasound GSoC students work on embeddings, Github actions + OOI data, notifications, and v3 of Orcasound web app
• May: WGU hackathon explores human and machine detection databases (API for CosmoDB | SQL queries with Postgres DB)
• Apr: 1-day Microsoft hackathon generates PRs to ai4orcas-livesystem to improve OrcaHello live inference system; HALLO project developing approaches to ecotype and pod classification
• Mar: After 6 months of beta-testing, Microsoft real-time inference system reaches 1300 detections (200 true positives), implements moderator UI improvements; HALLO project generates new labeled Canadian data; Akoustos project spins up.
• Feb: Earth Species project adds Orcasound data to their library
• Jan: Val/Scott presents Jetson Nano "edge computing" talk at Meridian Winter Webinar; 3 project pages published at ai4orcas.net

2020

• Dec: 2020 GSoC team presents at ASA; GSoC students present Orca Active Learning talk at Merdian "Winter Webinar" & Acoustical Society of America virtual meeting; Microsoft lead devs present Real-time "ML in the Wild" at Meridian "Winter Webinar"
• Nov: 2020 GSoC team hack re OrcaAL & AK KWs; live inference running beta-test on all 3 nodes.
• Oct: Val develops and tests edge detector on Jetson nano at Orcasound Lab; Microsoft team labels round 10 of Orcasound data, retrains models.
• Sep: Microsoft's AI for Orcas team deploys real-time inference system on Orcasound Lab hydrophone for testing of performance and human moderation
• Aug: Google Summer of Code (GSoC) students Kunal and Diego develop an active learning tool (OrcaAL app | OrcaAL research)
• Jul: Microsoft annual hackathon (7/27-29) goal: real-time inference prototype! DemocracyLab Create-a-thon (7/18, focus on design/dev of human learning UI)
• Jun: AI4orcas project page launched, including Google Summer of Code project on active learning
• May: DemocracyLab Hack for Your Mother (5/9, Orcasound 2.0 web app launch)
• Apr: Orcasound earns $15k Azure credits with University of Washington for the "Detect2Protect" proposal to the Leonardo DiCaprio Innovation Grant via Microsoft's AI for Earth program
• Mar: DemocracyLab Hacky-St.-Patrick's-Day (3/14, Orcasound); Microsoft remote-only hackathon (3/28, Pod.Cast + OrcaHello)
• Feb: Microsoft Hackathon (2/29, Pod.Cast + OrcaHello); Orcasound becomes host organization for Google Summer of Code
• Jan: Hackathons at Democracy Lab (1/11, Orcasound) and Microsoft (1/25, Pod.Cast + OrcaHello)

Recent progress (2019-2020)

2019

• Nov: Hackathons at Microsoft (11/7), UW (11/16), and DemocracyLab (11/23); Orcasound presents at Canadian deep learning and bioacoustic metadata workshops (11/19-22)
• Oct: Organize train/test data with Orcasound members, data, and Microsoft labeling tool (Pod.Cast)
• Sep: Submit UW-led ML proposal to Leonardo Dicaprio Foundation + Microsoft "Innovation Grant"
• Aug: Launch beta-testing of Orcasound UI with "I hear something" button for human listeners
• Jul: Microsoft annual hackathon (7/27-29): built "Podcast" annotation tool (Akash, Prakruti, & Nithya)

For more details, see the growing list of documentation pages for each Orcasound machine learning effort.

Deeper history of AI for Orcas project

Starting in the early 2000s, members of the Orcasound community have been contemplating the application of artificial intelligence to the problem of detecting orcas acoustically. Orcasound's AI for Orcas project page describes the evolution of our collective efforts. #ai4orcas