Machine Learning Image Recognition Analyst Needed

Integrated Statistics is looking for a Machine Learning Image Recognition Analyst to support the National Oceanic and Atmospheric Administration (NOAA) National Marine Fisheries Service (NMFS) Northeast Fisheries Science Center (NEFSC). The NEFSC Protected Species Branch (PSB) mission includes evaluating threats to species and populations due to anthropogenic activities. The Machine Learning Image Recognition Analyst will be working on protected species bycatch, which is when protected animals get caught or entangled in fishing gear. Bycatch can negatively affect species such as dolphins, sea turtles, protected fish, and whales by harming animals, contributing to population declines, and impeding population recovery.

For this project, video data that was collected on commercial fishing vessels as part of a fisheries monitoring program is being reviewed so that bycatch events and species can be identified and further analyzed. Since such interactions are relatively rare events, these data are sparse and an automated review of the video can be a realistic and cost effective way to identify actual or probable protected species interactions.

The Machine Learning Image Recognition Analyst will develop analyses of image data to inform bycatch estimates and assessments of fisheries impacts on protected species. The Analyst’s ultimate goal is to develop an algorithm automating the detection and identification of protected species visualized on video from gillnet, bottom trawl, and other types of fishing gear. The Machine Learning Image Recognition Analyst will begin by curating and developing a protected species (bird, mammal, turtle) image library utilizing video and still image data sets. After that, the Analyst will explore approaches to automating image analysis, developing algorithms in support of machine learning applications and annotated training data sets.

Duties

Gaining familiarity with the data and processes, including image review, data analysis, image data set curation, preparing and processing data. Identifying, acquiring and/or generating, and characterizing data and annotations.

Investigating and determining suitable machine learning technologies and available open-source resources (software applications, coding, and curated datasets) for a data-sparse context.

Training and testing models.

Developing machine learning based video analysis; creating a system and workflow for conducting the automated image analysis.

Generating reports that use the image data to inform assessments of fisheries impacts on protected species.

Qualifications and Skills

BS in Computer Science, Data Science, Mathematics, or a closely related field.

Deep proficiency in at least one programming language relevant to machine learning and artificial intelligence for image recognition (VIAME, TensorFlow, Python, R, Matlab, C++, etc.)

Demonstrated skill and knowledge in applying artificial intelligence and machine learning techniques to image data for image recognition.

Experience in photogrammetry and information extraction using machine learning, deep neural nets for image recognition.

Project development and management skills are desirable. Statistical programming skills are desirable.

Coursework and/or exposure to biological data is desired, but not required. The skills needed are in programming, not biology.

US citizenship and presence. Ability to comply with EO 14042.

Excellent written and oral communication skills.

Track record of success in working remotely. Demonstrated ability to contribute timely deliverables and work collaboratively as part of a team.

The position is expected to be full-time for one year, the rate of pay depends on skills and experience, and the position is eligible for the benefits described on the Integrated Statistics website. The position will probably start remotely but with the goal of intermittent in-person work at the NEFSC lab in Woods Hole, MA.