# Foreseeing the Invisible: Amodal Reconstruction of Leaf Fossil Images

arXiv: https://arxiv.org/abs/2608.04423 · HTML: https://mlsanity.com/paper/2608.04423/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 22 of its 29 references that are in the map.
- First posted on arXiv 5 Aug 2026.

## Builds on (most cited first, 22 of 22)

- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation](https://mlsanity.com/paper/1606.04797/) (2016)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [Vision Transformers for Dense Prediction](https://mlsanity.com/paper/2103.13413/) (2021)
- [DINOv3](https://mlsanity.com/paper/2508.10104/) (2025)
- [SAM 3: Segment Anything with Concepts](https://mlsanity.com/paper/2511.16719/) (2025)
- [PVT v2: Improved Baselines with Pyramid Vision Transformer](https://mlsanity.com/paper/2106.13797/) (2021)
- [EVA-02: A Visual Representation for Neon Genesis](https://mlsanity.com/paper/2303.11331/) (2023)
- [Tversky loss function for image segmentation using 3D fully convolutional deep networks](https://mlsanity.com/paper/1706.05721/) (2017)
- [Semantic Amodal Segmentation](https://mlsanity.com/paper/1509.01329/) (2015)
- [Self-Supervised Scene De-occlusion](https://mlsanity.com/paper/2004.02788/) (2020)
- [MobileCLIP: Fast Image-Text Models through Multi-Modal Reinforced Training](https://mlsanity.com/paper/2311.17049/) (2023)
- [pix2gestalt: Amodal Segmentation by Synthesizing Wholes](https://mlsanity.com/paper/2401.14398/) (2024)
- [Amodal Ground Truth and Completion in the Wild](https://mlsanity.com/paper/2312.17247/) (2023)
- [Amodal Segmentation Based on Visible Region Segmentation and Shape Prior](https://mlsanity.com/paper/2012.05598/) (2020)
- [AISFormer: Amodal Instance Segmentation with Transformer](https://mlsanity.com/paper/2210.06323/) (2022)
- [Coarse-to-Fine Amodal Segmentation with Shape Prior](https://mlsanity.com/paper/2308.16825/) (2023)
- [Open-World Amodal Appearance Completion](https://mlsanity.com/paper/2411.13019/) (2024)
- [PLUG: Revisiting Amodal Segmentation with Foundation Model and Hierarchical Focus](https://mlsanity.com/paper/2405.16094/) (2024)
- [Amodal SAM: A Unified Amodal Segmentation Framework with Generalization](https://mlsanity.com/paper/2604.20748/) (2026)
- [Learnable Shape Prototypes with Occlusion-Geometry-Guided Injection for Amodal Instance Segmentation](https://mlsanity.com/paper/2605.24533/) (2026)

## Abstract (arXiv)

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.