# Active Learning for Cascaded Object Detection: Balancing Coverage and Uncertainty in Table Extraction Pipelines Source: [Active Learning for Cascaded Object Detection: Balancing Coverage and Uncertainty in Table Extraction Pipelines]() Collection index: [llms.txt]() Data updated: 2026-07-03T05:00:04.967Z Authors: Eliott Thomas, Mickael Coustaty, Aurelie Joseph, Gaspar Deloin, Vincent Poulain d'Andecy, Jean-Marc Ogier Published: 2026-07-01T10:30:05.000Z ## Abstract Table extraction from business documents relies on a cascaded pipeline where Table Detection (TD) first localizes tables and Table Structure Recognition (TSR) then recovers their internal layout. Building task-specific training sets for this pipeline is costly, particularly for TSR which requires fine-grained structural annotations. Active learning (AL) can reduce this annotation burden, yet most AL strategies are designed for single-model tasks and do not account for inter-stage dependencies in cascaded architectures. In this work, we present the first adaptation of Uncertainty Herding (UHerding), a hybrid coverage-uncertainty sampling method originally proposed for image classification, to cascaded object detection pipelines. We propose two pipeline-aware extensions that exploit the TD-to-TSR dependency: RankFusion adds dual-manifold coverage over both detection and structure representation spaces, while CAPA further incorporates stage-dependent gating and per-task uncertainty calibration. Extensive experiments across two public (PubTables-1M and FinTabNet) and two private table extraction datasets, with various annotation budgets (from 71 to 500 documents) show that UHerding generalizes well to table extraction, outperforming each baseline. Among pipeline-aware variants, RankFusion achieves higher expected gains but at the cost of greater variance, while CAPA emerges as the most consistent strategy, outperforming standard UHerding on three out of four datasets. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Active Learning for Cascaded Table Extraction on invoice-int](): Active Learning for Cascaded Table Extraction; invoice-int; AUC: 66.24. - [Active Learning for Cascaded Table Extraction on business-int](): Active Learning for Cascaded Table Extraction; business-int; AUC: 57.51. - [Active Learning for Cascaded Table Extraction on FinTabNet](): Active Learning for Cascaded Table Extraction; FinTabNet; AUC: 70.52. - [Active Learning for Cascaded Table Extraction on Cross-dataset Aggregate](): Active Learning for Cascaded Table Extraction; Cross-dataset Aggregate; Mean AUC: 65.74. - [Active Learning for Cascaded Table Extraction on PubTables-1M](): Active Learning for Cascaded Table Extraction; PubTables-1M; AUC: 70.86. ## Original sources and implementation [Original paper]() [PDF]()