Improved Airbus Pleiades trucks algo; a new & improved time series chart for results; temporary deprecation of alerts; and other bugfixes.
- Airbus Pleiades Trucks - updated model with improved performance. Now using pansharpened (RGB) imagery, and with pixel clouds enabled.
- Improved country/state classification for clusters falling just outside admin boundaries. Logic is now identical to that for AOIs.
- Bug fix for privacy filtering (was previously over-aggressive on filtering heatmaps)
- Bug fix to make results deterministic (if you run the same project multiple times, it now gives identical results)
Charts v2: beta release of charts v2 with improved performance & slicker user experience. Try this out by clicking “Switch to new chart” in the results view
Deprecated alerts/triggers from the UI. Will bring this back after we put in proper anomaly detection (tentative Q4 roadmap). Existing projects with alerts will continue to function as before.
- Faster load time, with updates to jupyterlab & plotly libraries
- go_utils API wrapper now supports skysat algos
- New tutorial notebook for seasonal analysis of hourly foot traffic projects
- Fixed regression on annotation causing bounding-box colors to appear grey when editing
- Fixed regression on AOI catalog preventing selection across multiple pages of AOIs
- Fixed window resizing issues on results page affecting map & chart
- Now reset foot traffic dwell time setting when switching back from hourly to daily observation frequency (reported by customers as confusing)