Related Experiment Video
Updated: Jul 11, 2025
05:41
A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
Published on: February 6, 2020
9.4K
DeepPatent2: A Large-Scale Benchmarking Corpus for Technical Drawing Understanding.
Kehinde Ajayi1, Xin Wei1, Martin Gryder1
1Computer Science, Old Dominion University, Norfolk, Virginia, 23529, US.
Scientific Data
|November 7, 2023
Summary
Researchers developed DeepPatent2, a large dataset of technical drawings, to improve computer vision tasks like image captioning and 3D reconstruction from sketches.
Area of Science:
- Computer Vision
- Natural Language Processing
- Machine Learning
- Data Science
Background:
- Computer vision (CV) and natural language processing (NLP) advances rely on large datasets.
- Current datasets limit CV tasks like image captioning on technical sketches and 3D reconstruction.
- Technical drawings in patents are an underutilized data source for CV research.
Purpose of the Study:
- Introduce DeepPatent2, a novel large-scale dataset of technical drawings.
- Address limitations in existing datasets for CV tasks involving scientific and technical imagery.
- Facilitate research in areas like conceptual captioning, 3D reconstruction, and image retrieval.
Main Methods:
- Extracted over 2.7 million technical drawings from US design patent documents (14 years).
- Annotated drawings with 132,890 object names and 22,394 unique viewpoints.
- Developed a dataset pipeline for large-scale data extraction and processing.
Main Results:
- DeepPatent2 comprises 2.7 million technical drawings with rich metadata.
- Demonstrated dataset's utility for conceptual captioning tasks.
- Showcased potential for advancing 3D reconstruction and image retrieval.
Conclusions:
- DeepPatent2 offers a valuable resource for CV and NLP research on technical imagery.
- The dataset can significantly improve performance on tasks requiring understanding of sketched and technical visuals.
- Future research can leverage DeepPatent2 for diverse applications in scientific and engineering domains.
Related Concept Videos
01:15Depth Perception and Spatial Vision
678
Depth perception is the ability to perceive objects three-dimensionally. It relies on two types of cues: binocular and monocular. Binocular cues depend on the combination of images from both eyes and how the eyes work together. Since the eyes are in slightly different positions, each eye captures a slightly different image. This disparity between images, known as binocular disparity, helps the brain interpret depth. When the brain compares these images, it determines the distance to an object.
678
01:12Modeling and Similitude
270
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
270
01:10Design Example: Measuring Distance Between Two Points with Obstructions
42
When measuring distances in areas with physical obstructions, such as a lake in a field, surveyors must employ techniques to calculate accurate lengths without direct line measurements. One effective method is the offset technique, which allows for precise distance estimation over inaccessible stretches.In this scenario, a surveyor must measure a side of an area that crosses a lake. Since the measuring tape cannot span the lake, the surveyor begins by establishing a baseline that aligns with...
42