PlantShade — Project Website

Static project page for PlantShade: Predicting Plant Shadows for Lighting-Aware Robotic Agricultural Operation (IROS 2026). Adapted from the DeepShade × ShadeBench demo template, restructured around the PlantShade pipeline.

Structure

index.html          single-page site
styles.css          botanical dark/light theme (self-contained, no external fonts)
app.js              interactivity + real Table II data + NRH photosynthesis curve
prepare_assets.py   pulls & downscales real demo frames from dataset/plantAll
assets/             generated demo images (real dataset frames)

Sections (pipeline order)

1. Data collection — Helios–UE5 pipeline, dataset specs, release link
2. Preprocessing — (source, target, prompt) triplets, split
3. Interactive demo — circular supplementary-light slider (100 positions)
4. Model — ControlNet + SD2.1 architecture, training command
5. Inference — before/after compare (RGB ↔ shadow map)
6. Results — Table II (PlantShade vs Stable Diffusion), tabbed by layout, ID/OOD
7. Photosynthesis — interactive NRH light-response curves per species