Description

All core steps should use sv.Detections as internal structure to store detections + changes into EE to bring conditional eval capabilities.

Type of change

Please delete options that are not relevant.

• New feature (non-breaking change which adds functionality)

How has this change been tested, please provide a testcase or example of how you tested the change?

• unit tests + integration tests
• tested with load on staging (object-detection, classification and instance-segmentation lambdas) as usual - 🟢
• integration tests for platforms:
  • Jetson 4.6.1: https://github.com/roboflow/inference/actions/runs/9676683723/job/26696795212 🟢 on initial run, retrying to confirm there is no impact of issue with zxing-cpp
  • Tesla T4: https://github.com/roboflow/inference/actions/workflows/test.nvidia_t4.yml 🟢
  • Jetson 5.1 🔴 RELEASE BLOCKER - local build fails as described at Slack - the problem is probably with newest pybind11 release (added 6hrs ago 😠) which now defaulting for from-sources build of zxing-cpp with no way to override that already discovered. - RESOLVED, all tests 🟢
• run workflows prepared on testing session against staging 🟢

Any specific deployment considerations

N/A

Docs

TODO: Docs need to be updated