Progress 09/01/19 to 06/15/22

Outputs
Target Audience:Prior to shutting down the project, our commercialization plan was focused on selling Treetective to a broad range of target audiences, including municipal and state governments, environmental nonprofit groups, and community organizations. Changes/Problems:Azavea's Phase I street tree inventory prototype leveraged theGoogle Street Viewproduct to enable the efficient inventory of trees without having to do in-person field surveys. The Google Maps Street View offering was an ideal mix of affordable (commercial use is affordable and the cost scales linearly with usage), consistency (the imagery and positioning data is the same for every location), and quality (the positioning data is reliable and consistent through their API). This was critical to the success of this project. In our Phase II proposal, however, we noted a potential risk that Google Maps' terms and conditions might not allow ownership of data that derived from this process, or that they might change the terms and conditions to preclude its use completely. In an effort to mitigate the risk with Google Maps Street View, we identified two other services, Cyclomedia and Mapillary, that provide alternative services for street-level imagery in many cities. In our most recent interim status report, we also indicated that we had identified a potential third alternative, Microsoft Bing Maps. Unfortunately, after our Phase II award was received,we learned that Google Maps had changed its terms and conditions in two ways. First, they prohibited "creating data from Street View images, such as digitizing or tracing information from the imagery". Second, they added a specific stipulation forbidding use of Street View imagery for tree inventories. We do not know why they added this stipulation, but it completely prevents us from using Google Maps Street View for this project. This was a blow, but we initially believed that we had two (and possibly three) backups: Mapillary, Cyclomedia, and Microsoft Bing. We have explored each one in detail over the past year. Mapillarywas acquired by Facebook in June 2020, but it continues to operate. However, we found that the data is not captured with sufficient detail and, more importantly, consistency. Mapillary crowdsources its data from tens of thousands of contributors, who use an array of disparate hardware and conditions. We found that it was not possible to develop a general use software tool that would be able to support tree inventory from imagery collected in this manner. Cyclomediacollects data on a consistent hardware platform. However, they distribute the data as "point clouds" that would require a great deal of additional processing, as well as carrying a significant license fee. Pursuing this path would not only require a completely different technical work plan from the one submitted in our Phase II proposal, but also significantly raise the product's cost to consumers due to the license fee, most likely pricing it well out of their means. Microsoft Bing Mapsprovides a similar service to Google Maps Street View. It is not available in as many cities, but it is consistent, it is high quality, and it is provided through an API that is similar to Street View. However, after testing it, Microsoft provided us with a quote of $250,000 to implement the service just for this R&D work proposed in our grant application. This is not in our budget and is not a viable approach for a commercial product in the urban forestry space, as it would again price the product beyond the means of those most in need. Finding a suitable data provider has been our biggest impediment since the Phase II award, and we have tried every avenue we can think of to make it work under the current scope. We are aware of a potential antitrust action by the U.S. Department of Justice against Google for the restrictive terms in its Google Maps product. However, the Department of Justice has not yet taken any action, and there is no guarantee that will take any action in the future. Even if it did, it might take significanttime before we know if there is a remedy for the term restriction, and for the remedy to be implemented in a way that would make this project viable again. We are further aware that we have already been granted one no-cost-extension, and we are not eligible for another. Based on the above referenced legal requirements, data quality issues, and licensing constraints, we no longer believe our project can be technically or commercially viable, and we are unable to proceed with further development. What opportunities for training and professional development has the project provided?Investment in a stimulating, intellectually challenging, and open research environment for Azavea employees through opportunities such as the Small Business Innovation Research program is an explicit part of the company's culture, as well as being a strong attraction to potential job candidates and an element of our retention strategy. While we did not succeed in developing a commercial product, we learned a great deal about the remote tree inventory process through our consultant, Dr. Adam Berland. We also learned about the availability, consistency, quality, and licensing requirements associated with many online mapping services. How have the results been disseminated to communities of interest?Azavea was working with Adam Berland, Assistant Professor of Geography from Ball State University, as a consultant on the project. We have made him aware of our issues finding street-level imagery suitable for remote tree inventories, and that we have decided to cancel the project. We are grateful for the assistance he was able to provide to this point. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Azavea completed several tasks on our Phase II project, butwe faced some impediments during the last two years that affected our ability to complete the work as initially planned. First, the COVID-19 pandemic impacted our development progress due to reductions in staff capacity and the impact of the pandemic lockdown on several key work plan tasks. Azavea requested a one-year no-cost extension to this project to better position us to meet these challenges, and this extended the period of performance until August 31, 2022. In the meantime, we continued to grapple with an additional impediment that we first described in our Phase II proposal: access to street level imagery for a broad range of cities in North America. In short, we needed an affordable, consistent, and high-quality street level imagery data source with terms and conditions that would support its use in a commercial product. The Google Maps Street View offering that we leveraged in our Phase I prototype met these requirements. As we prepared our Phase II proposal, however, we became aware of a potential risk that Google Maps' terms and conditions might not allow ownership of the data that was derived from this process, or that they might change the terms and conditions to preclude its use. In an effort to mitigate this risk, we identified two potential alternatives prior to submitting our Phase II proposal, and a third since then. In the year following our Phase II award, Google Maps did indeed change its terms and conditions to such an extent that their Street View imagery may no longer be used in any tree inventory project. Even worse, the two potential imagery sources we had identified as alternatives, Cyclemedia and Mapillary, turned out to lack sufficient quality and consistency. A third that we identified after the beginning of our Phase II work, Microsoft Bing Maps, was subject to a significant license fee that made it commercially infeasible. As further described in this report, and as previously discussed with our USDA/NIFA program officer, we have concluded that our project cannot be technically or commercially viable in the urban forestry domain under these conditions, and we have found no further avenues available for us to explore. Therefore, we believe it is necessary to cancel our Phase II project with the submission of this report and to relinquish the remaining funds for use by other USDA SBIR recipients.

Publications

Progress 09/01/20 to 08/31/21

Outputs
Target Audience:As we enter Year 3, Azavea continues to work with Adam Berland,Assistant Professor of Geography from Ball State University, as a consultant on the project. We have met with Dr. Berland remotely to review the proposed wireframes and discuss updates to his research regarding tree inventories using ground-level imagery. Dr. Berland provided valuable feedback on wireframes that we integrated into the revised design prototype. Changes/Problems:When we accepted the Phase II award in September 2019, our projected workload for the next year indicated we would be able to complete all proposed tasks within the original two-year grant timeframe. Since the COVID-19 pandemic, however, our schedule has been impacted by reductions in staff capacity due to higher childcare burden, the opportunity to contribute to the COVID-19 response effort through a number of high priority projects with the World Bank and nonprofits, and the impact of the pandemic lockdown on several key work plan tasks. Azavea requested a one-year no-cost extension to this project to better position us to meet these challenges, and this has extended the period of performance until August 31, 2022. Gathering specific information on how we may use ground-level images in a commercial application has also proven challenging. Uncertainties regarding how we may access and charge for ground-level imagery impact our ability to implement options for users to select from multiple imagery options and may ultimately impact customer costs. What opportunities for training and professional development has the project provided?We have received valuable feedback from Dr. Berland on multiple aspects of the remote tree inventory process that hassignficantly enhanced our user workflow designs. How have the results been disseminated to communities of interest?The ongoing ramifications of the COVID-19 pandemic continue to impact our ability to disseminate Treetective morebroadlyto our public and private sector communities of interest. Some of our planned dissemination activities, such as attendance at trade shows and traveling to meet with potential clients, have not been possible due to ongoing travel and social distancing recommendations. Werecently hired a Marketing Lead that has been tasked withcontent marketing andsocial media outreach for Azavea's products and services, and we anticipate that this individual will assist with future dissemination activities for Treetective as new features and functionality are developed. What do you plan to do during the next reporting period to accomplish the goals?We anticipate increasing technical development on the project over the next several months, engaging student researchers to assist with pilot testing, and identifying at least one street-level imagery provider with licensing and fee options suitable for use in acommercial application.

Impacts
What was accomplished under these goals? Since our last interim report, we developed an in-depth series of wireframes and implemented design prototypes outlining the new functionality envisioned for the application. After establishing the backend architecture for Treetective, we then focused on implementing the administrative interfaces that are necessary to create and manage organization accounts, user authentication and registration workflows, and map interactions on the site's main pages. We also designed a user authentication system that supports users creating accounts, resetting passwords, tracking the number of trees inventoried by a user, and viewing progress on a survey area. We discussed various methods for defining survey areas for remote inventories, including selecting pre-defined geographic boundaries (i.e. zip codes, neighborhoods, etc.), but decided to support drawing the area on a map. This option supports the greatest flexibility in identifying locations, as administrators are not limited to a specific set of pre-set boundaries. In addition to completing these technical tasks, we extended our research into additional data providers and the terms of use connected to integrating their ground-level imagery into a commercial application.

Publications

Progress 09/01/19 to 08/31/20

Outputs
Target Audience:Azavea has remained in regular contact with our consultant, Dr. Berland, via video calls and emails, to ensure the product will meet the needs of our target audience ofmunicipal and state governments, environmental non-profit groups, and community organizations.He has provided feedback on the revised wireframes and counseled us regarding various imagery options. We have not yet reached a point in our development process where the application's new features are complete enough to support data gathering with our targer user community, but we had the unexpected opportunity to test the prototype with US Forest Service and other tree tenders in Philadelphia who were eager to complete their planned summer inventory remotely in the face of COVID-19 restrictions to in-person activities. Some of our planned commercialization activities, such as attendance at trade shows and traveling to meet with potential clients, will not be possible due to travel and social distancing recommendations related to COVID-19. We are currently working with our Business Development Manager to create a new plan that will incorporate content marketing, social media outreach, remote demonstrations, and introductory pilots that can be carried out regardless of thepandemic. Changes/Problems:When we accepted the Phase II award in September 2019, our projected workload for the next year indicated we would be able to complete all proposed tasks within the two- year time period. Since the COVID-19 pandemic, we have we have both taken on some unplanned projects related to COVID response and have had some reductions in worker capacity due to a higher childcare burden for most workers with children. We have taken steps to hire additional engineers to make up the difference, but hiring processes have taken longer than usual due to the remote nature of the hiring process. This increased project workload and reduced capacity has prevented us from completing the amount of development work we had initially planned for the past six months. In order to maintain progress on the technical objectives, we focused on developing the user interface and system architecture design in the first year. The refinement and specificity of these designs will assist in quickly implementing the features in the coming months. We anticipate increasing technical development on the project in fall 2020, particularly as we have now successfully hired an additional software engineer on the team assigned to the Treetective project. Gathering specific information on how we may use ground-level images in a commercial application has also proven challenging.We continue to research the licensing and fees associated with the various ground-level imagery options, with the goal of implementing multiple imagery selections in the application. What opportunities for training and professional development has the project provided?In the first half of the grant period, we researched user needs and developed an in-depth series of wireframes and design prototypes outlining the new functionality we intend to develop. How have the results been disseminated to communities of interest?The ongoing ramifications of the COVID-19 pandemic have both positively and negatively affected our ability to disseminate Treetective to our communities of interest.First, on the negative side, many organizations have experienced significant negative impacts to their budgets. Municipal governments face budget shortages and many non-profit organizations have had to furlough or lay off employees due to declines in revenue. Budget shortages may decrease the likelihood that government and non-profit groups can acquire new technology such as Treetective. Grant funding to support non-profit programs may also be limited, or the funds redirected to pandemic-related programming rather than technology budgets. We expect we will need to reconsider potential price points, subscription lengths, or other budget related components of selling Treetective. Second, on the positive side, the wide-spread adoption of social distancing and remote work protocols has increased the need for tools such as Treetective. When organizations may be unable to physically send groups of surveyors into the field to identify and measure trees, Treetective enables them to continue tree surveys using an online tool. .For example, we were approached by the US Forest Service Philadelphia Field Station and the Pennsylvania Horticultural Society this spring regarding potential use of Treetective for their planned summer tree inventory. We provided them with access to the prototype application, and they were able to continue their inventory despite in-person survey restrictions due to the pandemic. We expect organizations will continue to limit physical tree inventories, particularly those conducted by interns or volunteers, until a COVID-19 vaccine is widely available. We will adjust our outreach plans to emphasize how Treetective can assist groups in conducting remote inventories when in-person inventories are not an option. What do you plan to do during the next reporting period to accomplish the goals?We hope to begin the process of hiring student researchers and conducting pilot remote inventories in the next few months as the new features in the application are rolled out to production. The student researchers will be asked to conductsample inventoriesand canopy assessments in order to gauge the effectiveness of remote inventories, particularly in terms of genus/species identification, trunk diameter measurement, and canopy measurement.

Impacts
What was accomplished under these goals? In the first half of the grant period, we researched methods for calculating tree canopy on an individual tree basis and developed wireframes outlining our selected option. We propose a process where the Treetective user first triangulates the location of the tree by clicking on it in two different images. They enter basic information about the tree (species, condition, etc.) and use a measurement tool to calculate the tree's diameter. We then switch to a top-down satellite image view with an icon placed at the center of the tree's canopy based on the location identified for the tree's trunk. The screen shows a second caliper measurement tool, which the user adjusts to measure the diameter of the canopy. We include the resulting diameter in the data that administrators can export from Treetective for use in other applications. Azavea also researched user needs and developed an in-depth series of wireframes and design prototypes outlining the other new functionality planned for the application. We reviewed these wireframes with Professor Berland, Assistant Professor of Geography from Ball State University and our consultant for this project, and adjusted the designs based on hisfeedback. These design prototypes will guide development of the features in the second half of the grant period. We also began our research into additional data providers and the terms of use connected to integrating their ground-level imagery.

Publications