GTR (Gated Token Recurrence) for LibreYOLO: detection, instance segmentation, pose, oriented boxes, depth and semantic segmentation. MIT.
AI & ML interests
None defined yet.
Recent Activity
LingBot-Vision self-supervised ViT backbones (Apache-2.0) with LibreYOLO-trained dense heads for ADE20K semantic segmentation.
EoMT (Encoder-only Mask Transformer, DINOv2 backbone) semantic and instance segmentation weights for LibreYOLO.
Image-classification model families packaged for LibreYOLO (ImageNet-1k, Apache-2.0). MobileNetV4-conv and ConvNeXt V1.
RF-DETR segmentation models (Nano, Small, Medium, Large) for LibreYOLO. Based on the RF-DETR architecture with a MaskDINO-inspired segmentation head.
Roboflow 100 datasets for training and evaluation
Real-time PP-LiteSeg semantic segmentation, Cityscapes 19-class, native rectangular canvases. NON-COMMERCIAL weights (Cityscapes terms).
Promptable segmentation (LibreSAM tier): point and box prompts, encode once and prompt many. Image inference; video is out of scope in v1.
COCO-pretrained D-FINE-seg weights (Apache-2.0, converted from ArgoSA/D-FINE-seg) for LibreYOLO task=segment.
Models released with the LibreYOLO use-cases repo (https://github.com/LibreYOLO/use-cases).
All LibreYOLO pretrained model weights
COCO dataset subsets for testing and validation
GTR (Gated Token Recurrence) for LibreYOLO: detection, instance segmentation, pose, oriented boxes, depth and semantic segmentation. MIT.
Real-time PP-LiteSeg semantic segmentation, Cityscapes 19-class, native rectangular canvases. NON-COMMERCIAL weights (Cityscapes terms).
LingBot-Vision self-supervised ViT backbones (Apache-2.0) with LibreYOLO-trained dense heads for ADE20K semantic segmentation.
Promptable segmentation (LibreSAM tier): point and box prompts, encode once and prompt many. Image inference; video is out of scope in v1.
EoMT (Encoder-only Mask Transformer, DINOv2 backbone) semantic and instance segmentation weights for LibreYOLO.
COCO-pretrained D-FINE-seg weights (Apache-2.0, converted from ArgoSA/D-FINE-seg) for LibreYOLO task=segment.
Image-classification model families packaged for LibreYOLO (ImageNet-1k, Apache-2.0). MobileNetV4-conv and ConvNeXt V1.
Models released with the LibreYOLO use-cases repo (https://github.com/LibreYOLO/use-cases).
RF-DETR segmentation models (Nano, Small, Medium, Large) for LibreYOLO. Based on the RF-DETR architecture with a MaskDINO-inspired segmentation head.
All LibreYOLO pretrained model weights
Roboflow 100 datasets for training and evaluation
COCO dataset subsets for testing and validation