Computer vision has recently made rapid progress, achieving a level of performance that was unexpected just a few years ago. This technology has opened possibilities in many real-world domains, including transportation, entertainment and safety. While these applications give value to our technology, research to date has predominantly focused on a few geographic regions, primarily the United States and Europe, raising concerns of globally unrepresented datasets, tasks and ultimately the direction of the field.
Parallel to this, internet penetration and mobile phone usage has dramatically increased throughout the world. Simple feature phones, often with cameras, are particularly pervasive in low- and middle-income countries, providing new entry points to address long-standing development challenges in areas like health, agriculture and education. The computer vision community could aid these efforts, but complex technical challenges prevent progress.
Facebook is calling for proposals for pilot and early-stage research that extends computer vision technologies in developing countries. We specifically seek projects that address the technical challenges impeding computer vision in these contexts, including data and hardware limitations and better integration of new information sources, such as high-resolution satellite imagery. Competitive applications will similarly leverage computer vision to achieve global development priorities, especially those captured in the United Nations Sustainable Development Goals. Illustrative computer vision applications for delivering this type of social impact include (but are not limited to) the following:
- Agricultural extension: better informing smallholders about crop health, pest control, livestock illnesses, or soil fertility
- Health services: improving diagnosis, dietary, water or sanitation assessments
- Education quality: tools to enhance access to information without requiring high levels of literacy or language skills
- Infrastructure and sustainable industry: improving factory machinery or training to increase safety, jobs or output. Transportation innovations for environmental and human benefit.
- Disaster relief and climate action: visual information for first-responders to fires, floods or other natural disasters. Technologies to manage or mitigate environmental change.
Facebook aims to support projects that align with our mission, past research, open source tools and state-of-the-art algorithms. Awards will be made in amounts ranging from $20,000 to $40,000 for projects up to 6 months in duration. Our goal is to support the implementation of computer vision applications, which require further testing, pilot data and/or partnership development. Applicants requesting higher budget amounts, however, will need to demonstrate more robust research designs, novelty and potential to achieve impact. No Facebook data will be provided to award recipients.
We encourage proposals from teams that include institutions, researchers and non-profits in developing regions.
 For the purposes of this RFP, we define “developing countries” as the low and middle-income economies designated by the World Bank.
Proposals should include
- A summary of the project (1-2 pages) explaining the area of focus, conceptual and methodological distinction between the pilot study and any future follow-on studies, a description of techniques, the research’s applicability to Facebook, a timeline with milestones and expected outcomes
- A budget description (1 page) including an approximate cost of the award and explanation of how funds would be spent
- Curriculum Vitae for all key project participants
- Organization details; this will include tax information and administrative contact details
Submission Deadline: Monday, April 29, 2019 at 5:00 pm PST
Illustration Photo: Cassava Brown Streak Disease symptoms, Tanzania. PlantVillage Nuru is an mobile app that uses Google’s Tensorflow machine learning tool and a database of images collected by crop disease experts across the world to diagnose crop disease. (credits: H.Holmes / RTB / Flickr Creative Commons Attribution 2.0 Generic (CC BY 2.0))