v0.2.0-beta

What's Changed

• Removed GitHub version of luxonis-ml from requirements-dev.txt by @kozlov721 in #109
• Add export output names specification for individual nodes by @sokovninn in #112
• Add support for multi-label visualization by @JSabadin in #110
• SequentialLR Scheduler by @JSabadin in #111
• Add OHEM losses by @sokovninn in #117
• Add model EMA (Exponential Moving Average) by @sokovninn in #114
• Add dynamic aliases for predefined models by @sokovninn in #118
• Mixed Precision Training by @JSabadin in #113
• Add gradient visualization during validation by @sokovninn in #115
• Add tests action with a manual trigger by @sokovninn in #123
• Fix tests action by @sokovninn in #124
• Force reinstall of luxonis-ml in tests action by @sokovninn in #126
• Config smart auto populate by @JSabadin in #127
• Updated docs action by @kozlov721 in #120
• feat: add n_validation_batches option to trainer by @JSabadin in #125
• Updated how we access config version from luxonis-ml by @klemen1999 in #122
• Unsupervised Anomaly Detection by @JSabadin in #116
• fix: call tracker.close() by @JSabadin in #119
• Run inference on the same device as used in the Trainer by @sokovninn in #121
• Pretrained weights for detection-light and segmentation-light by @klemen1999 in #128
• add test for cli commands by @JSabadin in #129
• added DDRNet to archive by @kozlov721 in #130
• Release v0.2.0 by @klemen1999 in #131

Full Changelog: v0.1.0-beta...v0.2.0-beta

v0.1.0-beta

What's Changed - Highlights

• Full support for multi-input and multi-task models
  • Multi Input Support by @CaptainTrojan in #36
  • Full Custom Task Support by @kozlov721 in #53
• Add video inference by @sokovninn in #84
• New Keypoint Heads and Losses by @JSabadin in #40
• Generate NN archive from training configs by @klemen1999 in #17
• Tuning - Augmentation Subsets Support by @jkbmrz in #35
• Train-Only Heads by @JSabadin in #83
• Code Cleanup and Improved Tests by @kozlov721in #69
• Tensor Core Float16 Precision by @kozlov721in #24