TAPe+ML: A Compact Structured Representation for Multi-Task Computer Vision

• 类型：arxiv
• 标识：2609.20869
• 链接：https://arxiv.org/abs/2609.20869
• 主分类：multimodal
• 形态：method
• TLDR：We present TAPe+ML v3, a compact computer vision system based on TAPe (Theory of Active Perception), a structured representation that encodes relations among perceptual elements before recognition. Instead of operating directly on pixel tensors, the system uses a shared TAPe representation and a modular recognition architecture for image classification, object detection, and instance segmentation. TAPe+ML v3 combines background and contour processing, local object localization, prototype-based classification, and a coordinator for specialized submodels. Across the reported experiments, it uses
• 待LLM分类：否
• 标题中文：TAPe+ML：面向多任务计算机视觉的紧凑结构化表示
• TLDR中文：我们提出 TAPe+ML v3，一个基于 TAPe（主动感知理论）的紧凑计算机视觉系统，该结构化表示在识别前编码感知元素之间的关系。系统不直接作用于像素张量，而是使用共享的 TAPe 表示与模块化识别架构，支撑图像分类、目标检测与实例分割。TAPe+ML v3 结合背景与轮廓处理、局部目标定位、原型分类以及专用子模型协调器。在所报告的实验中，该系统使用
• 来源文件：
• /inbox/tom/_candidates/2026-09-23-agent-rag-longcontext-candidates.json