Add state-of-the-art Machine Learning labels for fast review of bookmarked events—not days of scrubbing
The video importer expedites trip review by E-Monitoring Analysts. The labels generated by AI are organized by timestamp and camera to make it simple to import them as bookmarks into existing review tools.
The pipeline is extensible with Hooks to call any labeling services, and the Search API used to export bookmarks can also be used for fast ad hoc searches like [tuna and marlin].
E-Monitoring Analysts jump to the relevant video events rather than scrubbing through hours of video. Video events are labeled with the latest cloud Deep Learning classifiers without uploading Gigabytes of video.
import_video.py
from any laptop/computer.download_labels.py
to export the labels.The automated labeling runs on any Mac, Linux, or Windows computer. Since the pipeline is configurable, there are lots of options; so the simplest way to get up and running is to:
job_completed
that knows how to download the bookmarks and import them into your review tool upon job completion.
Camio is $0.04 per hour of imported video. For example, one camera on a 14-day fishing trip with continuously recorded video is $13.44 (14 days * 24 hours/day * $0.04/hour). This includes a) elimination of spurious motion events triggered by lighting changes, shadows, and ocean waves and b) labeling of humans, direction of movement, zone intersections. So [humans at fish door] works even with this basic service prior to advanced labeling.
The price of advanced labeling varies by choice of labeling services. However, Camio radically reduces the cost of advanced labeling by selecting only the salient, representative snapshot images from events to be submitted for labeling. The estimated cost of the fish labelers is $0.02 per 30-second video event. Contact us for details.