Object Detection Using Multiple Likelihood Maps with Weight Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection methods in images struggle to accurately recognize multiple objects of the same type that are close to each other, often resulting in erroneous detection or failure to detect individual objects.
Innovation Solution
The approach involves generating multiple likelihood maps with different weight parameters, using loss functions to adjust and update these maps, and integrating them to accurately detect the position of objects, even when they are adjacent or overlapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single likelihood map is used for object detection, then the detection process is simple and fast, but multiple objects of the same type close to each other cannot be separately recognized
Solution Approach 1:
The patent divides a single likelihood map into multiple likelihood maps, each responsible for detecting objects at different spatial locations. By segmenting the detection task across multiple maps with different weight parameters, the system can distinguish between closely spaced objects of the same type, thereby improving detection precision without requiring a fundamentally more complex detection framework
Solution Approach 2:
Each likelihood map is assigned different weight parameters that give it specialized sensitivity to specific spatial regions or object characteristics. This local quality differentiation allows each map to optimize for detecting objects in particular areas or under specific conditions, enabling better separation of closely positioned objects while keeping individual maps relatively simple
2Measurement precision
If multiple likelihood maps with different weight parameters are used, then multiple close objects can be separately detected, but the training complexity and computational load increase
Solution Approach 1:
The patent performs preliminary action by pre-defining multiple likelihood maps with different weight parameters before the actual detection task. These maps are trained in advance on training images containing multiple objects, allowing the system to learn appropriate weight configurations beforehand. This preliminary training enables the maps to be ready for rapid deployment during actual object detection without requiring complex real-time adjustments
Solution Approach 2:
The patent utilizes parameter changes by varying the weight parameters across different likelihood maps. During training, these parameters are adjusted to optimize detection performance for multiple objects. By systematically changing and optimizing parameters across multiple maps rather than using a single map with complex parameters, the system achieves better detection precision while distributing the computational burden across simpler individual maps
Data Source
AI summary
An apparatus includes a first acquisition unit that acquires a plurality of likelihood maps by setting a plurality of different weight parameters in a trained model that outputs, with an image feature extracted from the input image as an input, a likelihood map including, in association with a position in the input image, a likelihood indicating a possibility that the object is present, and a detection unit that detects, based on the acquired plurality of the likelihood maps, the position of the object included in the input image, wherein the trained model is a model that has learned the weight parameters based on loss values at least acquired using a first loss function for reducing a likelihood around a position of interest in the likelihood map, and a second loss function for increasing a likelihood acquired at the position of the object in the input image.