Building a litter-picking robot.
Powering the AI behind the litter survey "Takanome" with large-scale data
Pirika is an environmental startup on a mission to solve environmental problems with the power of science and technology. For the AI behind its litter-survey service "Takanome," ANOSUPO handled the annotation that turns the type, count and location of litter into data.
In-house work alone couldn't keep up with ever-growing video data
Takanome's AI needed large volumes of training data capturing the type, count and location of litter. But in-house work capped out at around 1,000 items a month — unable to keep pace with the video data piling up every month. "Data processing at a volume in-house couldn't handle" had become a bottleneck stalling development.
Pirika Inc., AI Development
A million items? With the way we worked back then, we still wouldn't have reached that today.
A Slack-centered workflow that minimized meetings and scaled throughput
ANOSUPO took on the image annotation Takanome's AI development required, building an operation centered on Slack communication that kept meetings to a minimum. By reducing friction while holding quality, throughput scaled to volumes in-house work could never reach.
Pirika Inc., Business Direction
On cost, about half. Completely different.
Cost ~1/2, processing to 100K/month, 1M+ items delivered
After adoption, annotation costs fell to roughly half. Processing scaled to 100,000 items a month (the previous vendor handled a few thousand), and cumulative delivered data passed one million items. A data foundation at a scale "we still wouldn't have reached the old way" now accelerates development toward a litter-picking robot.
Services used
The services we provided for this project.
Image & video annotationObject detection, segmentation, pose estimation and video tracking — pixel-level ground truth for computer vision.annotation-support.com
Ready to achieve the same results?
Even if your requirements aren't finalized yet, that's fine. Feel free to reach out.
