Skip to content

Releases: jonasrenault/adomvi

v1.3.0

09 Jul 08:34
Compare
Choose a tag to compare
  • Refactor notebooks to split dataset preparation, training and tracking
  • Use ultralytics' tracking algo instead of yolo-tracking
  • Remove Google colab notebooks
  • Add more datasets for training and evaluation

v1.2.0 - Google Image dataset

03 Aug 11:50
Compare
Choose a tag to compare
  • Fix google scraping utils
  • Create a dataset of annotated military vehicle images with four classes: AFV, APC, MEV & LAV

Dreambooth finetuning for tank image generation

26 Jul 14:27
0ea8ef4
Compare
Choose a tag to compare
  • Add a notebook to train dreambooth for generating tank images from a text prompt
  • Release trained dreambooth model weights

v1.0.0 - First models using public datasets

12 Jun 15:06
Compare
Choose a tag to compare

First version of adomvi with trained models (YOLOv8 nano and large) using ImageNet and CommonImage datasets.
Only one class in models: tank