Unified Object and Key Point Detection Using Shared Features

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for key point detection in images require two independent stages, leading to increased time and reduced detection speed when dealing with large numbers of objects, as each detected object needs to undergo separate feature extraction and key point detection, making the process less timely and practical for computer vision tasks.

Innovation Solution

A detection apparatus and method that integrates key point detection into the object detection process by using pre-obtained key point sets and shared features extracted from the image, allowing simultaneous feature extraction for both object and key point detection, independent of the number of objects in the image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If key point detection is performed in a separate stage for each detected object, then detection accuracy can be maintained, but detection speed decreases significantly when the number of objects is large

Engineering Contradiction:
Improvekey point detection accuracyVSAvoiddetection speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent merges the key point detection process with the object detection process into a single unified stage. Instead of performing key point detection separately for each detected object, the system performs both detections simultaneously using shared feature extraction, thereby maintaining accuracy while significantly improving detection speed when multiple objects are present

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a multi-functional detection system where a single detection stage performs both object detection and key point detection functions. The shared feature extraction mechanism serves both purposes simultaneously, making the system universal in handling multiple detection tasks without requiring separate processing stages

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If two independent detection stages are used (object detection and key point detection), then comprehensive detection can be achieved, but time consumption increases

Engineering Contradiction:
Improvedetection completenessVSAvoidtime spent in key point detecting operation
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent combines two independent detection stages into one unified detection process. By integrating object detection and key point detection into a single stage with shared feature extraction, the system achieves comprehensive detection coverage while eliminating the sequential time consumption of separate stages

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent performs feature extraction in advance as a preliminary action that serves both object detection and key point detection purposes. This pre-extracted feature set is then reused in the same detection stage for both tasks, avoiding redundant feature extraction and reducing overall time consumption

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS11393186B2Apparatus and method for detecting objects using key point sets
Publication Date: 2022.07.19 CANON KK
  • US11393186B2 patent drawing
  • US11393186B2 patent drawing
  • US11393186B2 patent drawing

AI summary

The present disclosure provides a detection apparatus and method, and image processing apparatus and system. The detection apparatus extracts features from an image, detects objects in the image based on the extracted features; and detects key points of the detected objects based on the extracted features, the detected objects and a pre-obtained key point sets. According to the present disclosure, the whole detection speed can be ensured not to be influenced by the number of objects in the image to be detected while the objects and key points thereof are detected, so as to better meet the requirements of timeliness and practicability of the detection by the actual computer vision task.