logo

Releases to Test Notes for 2020

July 20, 2020 Project Updates 3 minutes, 25 seconds

This article documents the major changes for the Cacophony Project's software updates each week. The latest updates are at the top of the article.

Week of 13th July

Browse Portal

Display tags and tag images properly

Thermal Processing

  • Added ability to ignore tags
  • Added keras model support

Week of 28th June

API Server

  • Make usernames, group names and devicenames case insensitive
  • Added visits API end point
  • Make Device name + Groups Unique
  • Added reporting for event service issues

Browse Portal

  • Added csv export to the visits page
  • Removed limit of number of visits and added load more visits button
  • Added sound lures to the visits page

Week of 16th March

Browse Portal

Bring back search button/panel on mobile and tablet views added support for multiple AI models

Thermal Processing

Added ability to run multiple models

Week of 2nd March

Browse Portal

Fixed search but where page wasn't being reset when search criteria changed.

Thermal camera

  • Rebuilt recorder (to incorporate raw writer which is only for use with bolson) - thermal-recorder(v2.5.0 -> v2.6.0)
  • Fixed bug so device config files are properly deleted when configuring a new device - device-register(v1.0.0 -> v1.1.0)
  • Updated modemd (this update was missed) - modemd (v1.0.0 - v1.1.1)
    • Rename setting and upgade go-config
    • Update loggind and require minimum connection time and max time between connections
    • Increase time between running lsusb to find modem to save energy
    • Reduce logging

Week of 24th February

API Server

NA

Browse Portal

NA

Thermal camera

  • Fix default SPI speed on camera

Week of 17th February

API Server

  • When pruning no longer required recordings from disk, wait 20 minutes between fetching the list and pruning objects. This should give the recordings time to be processed.

Week of 10th February

Thermal camera

  • Added API for getting and deleting events
  • Alerting for service failures

API Server

  • Avoid systemctl pager usage in postinstall script
  • Restored missing data from audio bait schedule API
  • Add systemError as tag type and update how JSONB are compared in sequelize

Browse Portal

  • Fix administration of groups

Week of 3 February

API Server

  • Fix prune objects (error in config was preventing from running)

Thermal camera

  • Show last time software was updated in About Page

Week of 27 January

No release

Week of 20 January

API Server

  • Normalise "false positive" tag
  • Fixed bulk tagging query returning archived tracks
  • Removed unused code
  • Remove some unnecessary logs
  • Restore GroupUsers for GET /api/v1/groups

Browse Portal

  • Tweaks for API changes
  • Suppress additional unnecessary api requests.
  • Adding support for bulk-tagging API
  • Allow anyone to view bulk tagger
  • Added tags for deer, sheep, and goat

Upload Processing

  • Cacophony Index fix for silent recordings: update correction curve so it is never negative

Thermal Camera

  • Change thermal camera socket and metadata handling to support new camera types
  • Show last timestamp update in management interface

Week of 13 January

Browse Portal

  • Tidy up the groups and device Admin pages
  • Added help for Visits feature
  • Fixed bug with page not loading for some URLs
  • Fixed bug with remembering preferred user playback speed

API Server

  • Reduce bug and improve code maintenance and navigation by using TypeScript
  • Add specialised API to speed up beta power tagging feature
  • Access tokens (JWT) support added to add/delete tag endpoints (for power tagging)
  • Access tokens for recordings are now only computed if the recording exists
  • Recording data is now deleted in the background at API call time, reducing the data that needs to be pruned later
  • Internal fileKey and rawFileKey fields are now longer exposed by the external API
  • Preserve API server Docker container after it stops so that logs are available for troubleshooting
  • Ensure all thrown errors are Error subclasses

Upload Processing

  • Pin to specific audio-analysis version to ensure consistent behaviour and support upgrades

Predator Classifier

  • Use config.toml for configuration (on Raspberry Pi)
  • Fixed bug where export would failed if clip was all preview
  • Fixed bug where closing the MPEG creater would fail if it was never used
  • Support snapshot generation (for management interface)
  • Fix time window calculation
  • Support motion without running the classifier (for Raspberry Pi)
  • Fixed invalid reference to configuration
  • Fix location tuple
  • Keep last frame in memory for management interface