SPECIAL NOTICE
R -- R--Scientific Development Services
- Notice Date
- 8/18/2021 6:50:08 AM
- Notice Type
- Special Notice
- NAICS
- 541620
— Environmental Consulting Services
- Contracting Office
- OFFICE OF ACQUISITON GRANTS SACRAMENTO CA 95819 USA
- ZIP Code
- 95819
- Solicitation Number
- 140G0321Q0196
- Response Due
- 8/25/2021 1:00:00 PM
- Archive Date
- 09/09/2021
- Point of Contact
- Allan-Loucks, Dana, Phone: 916-278-9344
- E-Mail Address
-
dallan-loucks@usgs.gov
(dallan-loucks@usgs.gov)
- Description
- Notice of Intent This is a notice of intent to solicit from a single source. The U.S. Geological Survey (USGS), intends to award a purchase order to Conservation Metrics, Inc, Santa Cruz, CA for unique artificial intelligence and machine learning software application as described in the attached Statement of Work. Award will be made in accordance with Federal Acquisition Regulation (FAR) 13.106-1(b) and the procedures at FAR Part 12 entitled, Acquisition of Commercial Items and FAR Part 13 entitled, Simplified Acquisition Procedures. Conservation Metrics, Inc is, to the USGS' knowledge, the only source who can provide a scalable pipeline for processing imagery. The purpose of this notice is to satisfy the requirements of FAR Subpart 5.2. This notice of intent is not a request for competitive quotes. However, all responsible parties may submit a quotation by the closing date of this announcement which shall be considered by the agency. No solicitation will be issued. A determination not to compete this action based upon responses received is in the sole discretion of the Government. The NAICS code for this action is 541620, Environmental Consulting Services. Responses or inquiries shall only be accepted through electronic mail addressed to dallan-loucks@usgs.gov and must be uploaded and received in their entirety no later than 08/25/2021 1500 ES. Responses submitted by hardcopy shall not be accepted or considered. Statement of Work: Scientific Development Services - Machine Learning Background: A range-wide census is conducted collaboratively each spring by the U.S. Geological Survey and others to monitor trends in abundance and distribution of the southern sea otter (Enhydra lutris nereis), also known as California sea otters, and thus to provide State and Federal resource agencies with the information requested for effective management for the recovery of this threatened species. Standardized censuses have been completed annually since 1982 and entails a combination of aerial and shore-based counts, providing an exhaustive count of the entire range of the sea otter in coastal California. In areas where shore-based counts are infeasible due to access or shore-based viewing limits, aerial surveys are flown along contiguous transects oriented parallel to the shore and covering all areas between the coastline and the 60-meter (m) depth contour. Standard protocols for these aerial surveys have involved sea otter biologists aboard a plane and visually identifying and counting sea otters in real time. In 2020, the annual sea otter census was cancelled due to COVID-related safety concerns. Future censuses are expected to be impacted by further logistical challenges (reduced availability and scheduling of aerial resources and personnel). To overcome some of these challenges, USGS is developing an aerial photographic survey of sea otters that would allow for fewer passengers on survey planes and simplify the logistics of sea otter surveys. It is critical that the new aerial photographic surveys yield information comparable to the traditional real time surveys, but there are unavoidable differences. In 2021, USGS conducted a pilot study for an aerial photographic survey of southern sea otters using methods that area already developed and being employed in a separate USGS study involving aerial photographs of seabirds. We collected approximately 28,000 high resolution photographs in the pilot study and now require machine processing to scan these images to identify and specifically count sea otters. To date, there has been very limited machine learning to detect sea otters in images due to limited target images that can be used to train the algorithm. We will need to use our collection of images to train the machine algorithm as well as generate counts. We will also need a process by which additional images collected in future surveys along these transects can be used to update the algorithm and improve the accuracy of sea otter counts. Inherent to these goals, we require the skills of experienced sea otter observers to verify true sea otter images and false images for the machine training and validation. Objectives: Develop an efficient process for counting sea otters in large quantities (tens of thousands) of aerial photographs, including machine training and optimizing performance accuracy, to be verified by skilled sea otter observers. This process should be updatable as additional aerial photographs are collected in future surveys. Tasks: 1. Develop and apply efficient processes for generating image datasets (verified targets and nontargets) for machine training and testing (validation) purposes. 2. Train a machine-based algorithm to use the training datasets from task 1 to detect sea otters in aerial survey images. 3. Quantify the numbers of sea otters detected and test the accuracy of this estimate using the validation datasets from task 1. Deliverables: 1. Training and validation test datasets. 2. Report of model development and performance. Guidance on repeating and updating the model using future photos. 3. Estimates of numbers of sea otters and degree of accuracy. Timeline: Work on the project will commence immediately upon award for the services. All work will be completed, and deliverables provided to USGS by Sep 30, 2022.
- Web Link
-
SAM.gov Permalink
(https://beta.sam.gov/opp/1984c50d9e644bb4b9a66dbbf26ba658/view)
- Record
- SN06101380-F 20210820/210818230118 (samdaily.us)
- Source
-
SAM.gov Link to This Notice
(may not be valid after Archive Date)
| FSG Index | This Issue's Index | Today's SAM Daily Index Page |