Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces AmodalDINO, a novel multi-head dense-prediction model designed for amodal reconstruction of leaf fossils, which predicts both visible and hidden leaf structures from a single RGB image without requiring a visible mask. By fine-tuning a DINOv3 ViT-L/16 model and incorporating auxiliary venation heads, AmodalDINO achieves impressive performance metrics, including 95.0% Dice and 90.5% IoU on synthetic leaf fossil images, and demonstrates strong transferability to real specimens. Additionally, the model's quantization to 4-bit weights allows it to operate offline in a browser while maintaining high accuracy, making it a practical tool for paleobotanists.
AmodalDINO can reconstruct the complete shape of fossil leaves from incomplete images, achieving state-of-the-art accuracy without requiring visible masks.
Fossil leaves are rarely preserved whole -- sedimentary rock hides, breaks, and erodes the lamina, yet paleobotany depends on the complete shape and outline of the leaf. We cast the recovery of the missing tissue as amodal reconstruction and present AmodalDINO, a multi-head dense-prediction model that predicts four masks from a single RGB image: visible leaf, amodal complete leaf, amodal main vein, and fine veins. Unlike essentially all prior amodal work, AmodalDINO is given no visible mask. It predicts the visible and amodal regions jointly, so it needs no upstream instance segmenter at runtime. Two simple but effective changes adapt the model to the amodal segmentation task: fully fine-tune a DINOv3 ViT-L/16 at a small learning rate instead of freezing it, and attach auxiliary venation heads alongside the leaf heads. These two changes enable the model to learn the structural shape prior of leaves. Trained only on synthetic leaf fossil images, AmodalDINO reaches 95.0% Dice / 90.5% IoU on the validation set and transfers well to real fossil specimens. Stripped to two heads, the same recipe can run on two benchmark datasets, reaching 85.05 full mIoU / 66.65 occluded mIoU on KINS and 80.90 / 38.15 on COCOA-cls. The model is also practical: by quantizing to 4-bit weights, it runs entirely offline in a browser, matching the original model with an IoU of 0.910. We also add ruler-based calibration to estimate surface area, and a generative visualization of living leaves on local devices.