Platform for distributed landcover feature data review
Methods and systems for extracting and reviewing landcover feature data are provided. An example method involves accessing geospatial imagery covering a geographic area and landcover feature data extracted from the geospatial imagery, subdividing the geospatial imagery into a grid of image tiles, assigning landcover feature data review technicians to review the image tiles and associated landcover feature data through a landcover feature data review platform, and receiving revised landcover feature data that modifies the landcover feature data.
1. A method comprising:
accessing geospatial imagery covering an area of interest and a corresponding set of landcover feature data that was extracted by a machine learning model with reference to the geospatial imagery covering the area of interest, wherein the area of interest is divided the area of interest into a grid of image tiles;
determining confidence data associated with each subset of landcover feature data within each image tile that reflects a level of confidence that the landcover feature data within the image tile was correctly extracted by the machine learning model;
generating and assigning tasks for team members of a team of landcover feature data review technicians to review the image tiles in which the landcover feature data is of low confidence;
providing functionality for the team members to view the geospatial imagery and landcover feature data within an assigned image tile and to make corrections to the landcover feature data with reference to the geospatial imagery; and
receiving one or more corrections to the landcover feature data from systems operated by the landcover feature data review technicians.
2. The method of claim 1 , wherein assigning the tasks comprises:
determining a unique priority value to be assigned to each image tile covering the area of interest, wherein the unique priority values are determined such that the priority value for each image tile is non-consecutive with each adjacent image tile; and
assigning tasks to team members based on an increasing or decreasing sequence through the priority values.
3. The method of claim 1 , further comprising:
providing functionality for the team members to review the corrections made by other team members and to generate tasks for team members to review previously reviewed image tiles.
4. The method of claim 1 , wherein each image tile comprises a working area and a surrounding buffer area, wherein the surrounding buffer area overlaps with the working areas of adjacent image tiles, and wherein the method further comprises:
providing functionality for the team members to generate a notification for a team member assigned to an adjacent image tile regarding an error, or correction, to landcover feature data within the buffer area.
5. The method of claim 1 , further comprising:
providing functionality for the team members to make corrections to elements of landcover feature data that span across two adjacent image tiles.
6. The method of claim 1 , further comprising:
fine-tuning the machine learning model based on the one or more modifications to the landcover feature data and the associated geospatial imagery.
7. The method of claim 1 , further comprising:
providing an online landcover feature data review platform with functionality to generate and assign the tasks to the team members, with functionality for the team members to make the corrections to the landcover feature data, and with functionality to receive the corrections to the landcover feature data.
8. The method of claim 1 , further comprising:
merging the one or more corrections to the landcover feature data with the landcover feature data.
9. The method of claim 1 , wherein the landcover feature data comprises vector data, and wherein the one or more corrections to the landcover feature data comprises a correction to the geometry of a polygon within the landcover feature data.
10. A system comprising one or more computing devices configured to:
access geospatial imagery covering an area of interest and a corresponding set of landcover feature data that was extracted by a machine learning model with reference to the geospatial imagery covering the area of interest, wherein the area of interest is divided the area of interest into a grid of image tiles;
determine confidence data associated with each subset of landcover feature data within each image tile that reflects a level of confidence that the landcover feature data within the image tile was correctly extracted by the machine learning model;
generate and assign tasks for team members of a team of landcover feature data review technicians to review the image tiles in which the landcover feature data is of low confidence;
provide functionality for the team members to view the geospatial imagery and landcover feature data within an assigned image tile and to make corrections to the landcover feature data with reference to the geospatial imagery; and
receive one or more corrections to the landcover feature data from systems operated by the landcover feature data review technicians.
11. At least one non-transitory machine-readable storage medium comprising instructions that when executed cause one or more processors to:
access geospatial imagery covering an area of interest and a corresponding set of landcover feature data that was extracted by a machine learning model with reference to the geospatial imagery covering the area of interest, wherein the area of interest is divided the area of interest into a grid of image tiles;
determine confidence data associated with each subset of landcover feature data within each image tile that reflects a level of confidence that the landcover feature data within the image tile was correctly extracted by the machine learning model;
generate and assign tasks for team members of a team of landcover feature data review technicians to review the image tiles in which the landcover feature data is of low confidence;
provide functionality for the team members to view the geospatial imagery and landcover feature data within an assigned image tile and to make corrections to the landcover feature data with reference to the geospatial imagery; and
receive one or more corrections to the landcover feature data from systems operated by the landcover feature data review technicians.