This paper reports the ICDAR 2026 competition on information extraction from atomic layer deposition and etching (ALD/E) scientific figures, built around the expert-annotated Sci-ImageMiner benchmark. The dataset covers four complementary end-to-end tasks. From 9 January to 8 April 2026, the competition attracted 68 active participants and received 1,263 public or private submissions. The reported results indicate that state-of-the-art multimodal models perform relatively well on classification and summarization, but remain weak on data extraction and scientific reasoning, especially visual question answering. The benchmark provides a domain-specific testbed for evaluating multimodal scientific figure understanding.
No heat snapshots are available in the last 24 hours.