Abstract:Scene understanding is central to general physical intelligence, and video is a primary modality for capturing both state and temporal dynamics of a scene. Yet understanding physical processes remains difficult, as models must combine object localization, hand-object interactions, relational parsing, temporal reasoning, and step-level procedural inference. Existing benchmarks usually evaluate these capabilities separately, limiting diagnosis of why models fail on procedural tasks. We introduce BARISTA, a densely annotated egocentric dataset and benchmark of 185 real-world coffee-preparation videos covering fully automatic, portafilter-based, and capsule-based workflows. BARISTA provides verified per-frame scene graphs linking persistent object identities to masks, tracks, boxes, attributes, typed relations, hand-object interactions, activities, and process steps. From these graphs, we derive zero-shot language-based tasks spanning phrase grounding, hand-object interaction recognition, referring, activity recognition, relation extraction, and temporal visual question answering. Experiments reveal strong variation across task families and no consistently dominant model family, positioning BARISTA as a challenging diagnostic benchmark for procedural video understanding. Code and dataset available at https://huggingface.co/datasets/ramblr/BARISTA.
Abstract:High-throughput 2D and 3D scanning electron microscopy, which relies on automation and dependable control algorithms, requires high image quality with minimal human intervention. Classical focus and astigmatism correction algorithms attempt to explicitly model image formation and subsequently aberration correction. Such models often require parameter adjustments by experts when deployed to new microscopes, challenging samples, or imaging conditions to prevent unstable convergence, making them hard to use in practice or unreliable. Here, we introduce DeepFocus, a purely data-driven method for aberration correction in scanning electron microscopy. DeepFocus works under very low signal-to-noise ratio conditions, reduces processing times by more than an order of magnitude compared to the state-of-the-art method, rapidly converges within a large aberration range, and is easily recalibrated to different microscopes or challenging samples.