We are excited to announce a new type of 'hackathon' we call an 'Open Innovation Challenge,' where teams will compete to find the most efficient solution using satellite imagery to detect invasive pests in LA County's tree canopy.
Individuals and unfunded teams can apply -- we are looking for cross-disciplinary teams: computer scientists, data scientists, geospatial scientists, entrepreneurs, designers, MLops, et al.
Teams will be provided pre-processed Landsat 8 data, GPU compute, and mentors to help them design and develop a proof of concept GeoAI-enabled pest infestation alert system in 4 weeks.
Info Session: March 8th at 5 pm PST Virtual
Kick-Off: March 18-20th in Los Angeles
Demo Day: April 21st in Los Angeles
It's mandatory that at least 1 team member participate in the in-person kick-off and demo day.
Individuals and unfunded teams can apply (ideally 3-5).
We are looking for cross-disciplinary teams: computer scientists, data scientists, geospatial scientists, entrepreneurs, designers, MLops, et al.
Teams will present on stage at AI LA's Earth Summit on April 21st at the BioscienceLA Collaboration Hub in Culver City.
The 3 winning teams will also receive office hours from Starburst Aerospace Accelerator and free corporate legal services from Wilson Sonsini, and other perks from our partners.
The judging criteria for this project reflect the need for hybrid design and data skills:
Please fill out this form to be considered as a participant in this challenge.
Trees provide many important benefits to the residents of LA County. In addition to beautifying our communities, they provide shade and cooling, support biodiversity, help manage stormwater, reduce air pollution, and improve mental health. Unfortunately, trees in the County face multiple threats. In addition to the ongoing threats of drought stress and heat stress, trees in the County also face the threat of invasive pests which have the potential to kill millions of trees in Southern California—of particular concern are the invasive shot hole borer (ISHB) and the golden spotted oak borer (GSOB). On May 18, 2021, the LA County Board of Supervisors adopted a motion, authored by Supervisors Kuehl and Solis, titled “Implementing an Early Detection Rapid Response Plan to Invasive Pests.” The motion directed the Chief Sustainability Office (CSO) to work with relevant County departments to develop a plan for early detection and rapid response (EDRR) to these invasive pests. One part of this plan is to pilot a remote sensing solution for early detection of invasive beetles.
County staff are racing against the clock to locate and control new infestations before they spread to surrounding areas. Today, the County relies on physical inspection, including surveying and trapping, to identify infestations of invasive pests. This boots-on-the-ground process is time-consuming, labor-intensive, and expensive, making it unrealistic for limited staff to manually survey every tree in the County. The County needs a way to frequently and comprehensively identify areas of potential new infestations Countywide so limited resources can be deployed to conduct physical inspections at priority locations. An automated remote sensing program using publicly available satellite data would provide an important monitoring tool to meet this need and help the County protect its vital tree resources.
This challenge gives participants the ability to consider the relationship between designing a targeted application and developing the data models utilized by that application.
Through this project, participants will explore building data models around a specific data challenge for the identification of trees with potential pest infestations. To successfully complete this challenge, teams must consider: What are we predicting and why? What is the end use for the data models we are training towards specifically?
Through research and conversations with stakeholders, teams go through a process of value discovery to identify applications that would be viable within the context of the stakeholder’s needs. This process of refining the technical goal must guide how teams utilize the data and identify relevant models.
Based on the identified end goal, participants may choose to utilize a computational, machine learning workflow, or a remote sensing spatial workflow to fulfill the data goals of this project. Workshops and available mentors will guide participants through possible workflows, although a combination of spatial and computational skills will be necessary to complete the final challenge.
This isn’t simply a data challenge. The goal of this program is to not only enable participants to learn data skills relevant to this use case, but to also give them the tools for R&D of product development with data.
February 22nd: Applications Open
March 11th: Applications Close
March 18-20th: Design and Development Workshop
Friday Evening: Stakeholder Presentation
Friday Evening: Teambuilding Activities
Saturday Morning: Design Workshop
Saturday Afternoon: Computational Analysis Workflow
Saturday Afternoon: Remote Sensing Spatial Analysis Workflow
Saturday Evening: Work Sessions
Sunday Morning: Work Sessions
Sunday Afternoon: Teams present their Design Breakdowns and projected Data Workflows to stakeholders for feedback
April 21st: Demo Day
User interface for multiple jurisdictions
Prospective analysis – Santa Monica Mountains Pilot
Proactive response (optional)
The lat/long and time stamps of the ISHB and GSOB infestations are available as pre-processed chips, with the data mapping to the Landsat time-series data.
During the March 18-20th workshop, we have a powerful data science workstation that can be on-prem during the kick off weekend.
Workstation Specs: Dual Socket Xeon GOLD 8 Core, 196gb DDR4 Ram, 1TB NVME SSD and Dual Nvidia Quadro RTX 8000
In the 4 weeks leading up to demo day, the data will have to be accessed remotely.
During the workshop weekend, participants will be offered a deeper dive into challenge considerations, as well as potential project workflows. Teams will use this information to present their proposed workflow and breakdown of design requirements to stakeholders at the end of the weekend for feedback.
The judging criteria for this project reflect the need for hybrid design and data skills. Judging criteria are: Data Workflow, Model Accuracy, Technical Implementation, UX Design, Alignment with Challenge, and Presentation.
For more information about this challenge, or if you have questions, please email OIS@joinai.la with the subject line, “OIC Question.” We will also be holding an information session with more information on March 8th.