Skip to content

Latest commit

 

History

History
95 lines (68 loc) · 3.72 KB

CHANGELOG.md

File metadata and controls

95 lines (68 loc) · 3.72 KB

Changelog

v1.7.3 (2023-09-16)

  • Addressed Issue #63 (#63)
  • Better version dependency pinning
  • Improved handling of image resolution and height/width dimensions

v1.7.2 (2023-05-04)

  • Upgraded geowombat (#72)

v1.7.1 (2023-05-04)

  • Pinned the upper version of pytorch-lightning and upgraded geowombat (#71)

v1.7.0 (2023-03-09)

  • (#49)
  • Improved and fixed issues in ResUNet 3+ Psi architecture, introduced in v1.6.5

v1.6.5 (2023-02-28)

  • Upgraded geowombat version following bug fix (#48)

v1.6.5b (2023-02-10)

  • Updated Python version in the Dockerfile (#45)

v1.6.4 (2023-02-01)

  • Pinned required packages to full semantic versions (#44)

v1.6.3 (2023-01-23)

  • Added chunksize argument for create-predict (#43)

v1.6.2 (2023-01-11)

  • Addressed security issue by bumping setuptools (#42)

v1.6.1 (2023-01-04)

  • Reorganized CLI arguments (#41)

v1.6.0 (2023-01-03)

  • (#40)
  • New architecture design based on UNet 3+ and residual convolutions
  • Modified total loss quantification with deep supervision of crop type in RNN layer
  • Added num_workers option in DataLoader for faster train/predict
  • Added .pt data compression by changing torch.save|load to joblib.dump|load

v1.5.2 (2022-11-10)

  • Fixed in-memory predictions (#38)

v1.5.1 (2022-11-10)

  • Fixes model outputs during serial predictions (#37)

v1.5.0 (2022-11-9)

  • PR #36
  • Re-configures the model architecture to improve the distance transform predictions
  • Improved prediction performance through pytorch-lightning batch predictions
  • Improves the edge calculations during cultionet create by using geometry boundary
  • Adds improved performance for mean and standard deviation calculations
  • Adds experimental orientation
  • Adds experimental Mask-RCNN
  • Adds a Docker file for CUDA 11.6

v1.4.0 (2022-10-24)

  • Added spatial k-fold cross-validation (#35)

v1.3.3 (2022-10-19)

  • Re-added serial predictions over CPU or GPU (#34)

v1.3.2 (2022-10-12)

  • Added default values for gain and offset

v1.3.1 (2022-10-12)

  • Added YAML files to distributed package data

v1.3.0 (2022-10-11)

  • Added a test dataset CLI option, additional model callbacks, and faster pre-model dimension checking (#33)

v1.2.3 (2022-10-04)

  • Added user argument to specify raster compression method (#32)

v1.2.2 (2022-10-03)

  • Updated geowombat version (#30)

v1.2.1 (2022-10-01)

  • Add concurrent predictions over image windows (#29)

v1.1.1 (2022-09-16)

  • Add argument for weight decay (#20)

v1.1.0 (2022-01-14)

  • Add config argument to specify region ids as a file (#28)

v1.0.1 (2022-06-18)

  • Fix band and time dimension getting with single-band images

v1.0.0 (2022-06-12)

  • Increase cultionet public version