Title: HCNQA: Enhancing 3D VQA with Hierarchical Concentration Narrowing Supervision Show Chinese title

Title: HCNQA：通过分层集中缩小监督增强3D VQA

Abstract: 3D Visual Question-Answering (3D VQA) is pivotal for models to perceive the physical world and perform spatial reasoning. Answer-centric supervision is a commonly used training method for 3D VQA models. Many models that utilize this strategy have achieved promising results in 3D VQA tasks. However, the answer-centric approach only supervises the final output of models and allows models to develop reasoning pathways freely. The absence of supervision on the reasoning pathway enables the potential for developing superficial shortcuts through common patterns in question-answer pairs. Moreover, although slow-thinking methods advance large language models, they suffer from underthinking. To address these issues, we propose \textbf{HCNQA}, a 3D VQA model leveraging a hierarchical concentration narrowing supervision method. By mimicking the human process of gradually focusing from a broad area to specific objects while searching for answers, our method guides the model to perform three phases of concentration narrowing through hierarchical supervision. By supervising key checkpoints on a general reasoning pathway, our method can ensure the development of a rational and effective reasoning pathway. Extensive experimental results demonstrate that our method can effectively ensure that the model develops a rational reasoning pathway and performs better. The code is available at https://github.com/JianuoZhu/HCNQA. Show Chinese abstract

Abstract: 三维视觉问答（3D VQA）对于模型感知物理世界和执行空间推理至关重要。答案导向的监督是一种常用于3D VQA模型的训练方法。许多采用此策略的模型在3D VQA任务中取得了有希望的结果。然而，答案导向的方法仅监督模型的最终输出，允许模型自由发展推理路径。推理路径缺乏监督使得通过问题-答案对中的常见模式发展表面捷径成为可能。此外，尽管慢思考方法推动了大型语言模型的发展，但它们存在思考不足的问题。为了解决这些问题，我们提出了\textbf{高容量神经网络问答}，一种利用分层集中缩小监督方法的3D VQA模型。通过模仿人类在寻找答案时从广泛区域逐渐聚焦到特定物体的过程，我们的方法通过分层监督引导模型进行三个阶段的集中缩小。通过监督一般推理路径上的关键检查点，我们的方法可以确保发展出合理且有效的推理路径。大量的实验结果表明，我们的方法可以有效确保模型发展出合理的推理路径并表现更好。代码可在https://github.com/JianuoZhu/HCNQA获取。