Sieve Aims to Give Your Raspberry Pi Security Camera Setup Cloud-Powered Machine Learning Smarts

Free for personal and student use, Sieve claims to be able to process "petabytes" of raw video into high-quality datasets.

Developer Mokshith Voodarla claims to have solved one of the biggest problems of using a Raspberry Pi single-board computer as a home security camera controller: sifting through hours of video to find frames of interest. His solution: Sieve.

"Sieve is a simple API that allows you to store, process, and automatically search your video data — instantly and efficiently," Voodarla explains. "Just think 10 cameras recording footage at 30 FPS, 24/7. That would be 27 million frames generated in a single day. Instead, with our API, you can take raw video footage that you captured on the edge in your deployment environments, upload it, query based on different metadata like motion, number of people, lighting, contrast, etc. and download a dataset of relevant labeled images."

The Sieve API aims to help you find interesting frames in hours of security camera footage. (📹: Sieve)

Designed to avoid the need to train a machine learning model on captured data, or to rely on a public dataset, the Sieve API offloads the hard work to the cloud — and needs only a few simple Python scripts to be installed on the Raspberry Pi to work, with the recommendation to use conda and the pip package management ecosystem for isolation and prerequisite installation.

Once installed and configured with an API key, using Sieve is three-stage process: First, the video is uploaded to the server; second, it's processed using a proprietary machine learning model; finally, the data appears in the Sieve Dashboard. From here, the user can pull individual frames based on a range of queries — from the number of people present in the image to the brightness of the lighting and even whether there are trees visible.

It's also free, Voodarla explains, though only for "personal use or if you're a student" — with pricing for commercial use available only on application. Its use of cloud computing, meanwhile, means that it's well-suited to relatively resource-constrained devices like the Raspberry Pi and other single-board computers — but does raise concerns about privacy.

More information on Sieve is available on the company website, where API keys can be requested, while sample Python scripts to upload video for analysis can be found on the project's GitHub repository.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles