Multi-Scale Object Inference for Real-Time Bounding Box Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine-learning models for computer vision in applications like autonomous vehicles face challenges in accurately detecting objects in real-time due to high computation time and flawed identifications, especially when objects are proximate or outside the receptive field, leading to unreliable decision-making.
Innovation Solution
The techniques involve training machine-learning models using improved hard example mining, non-maximum suppression replacement, and multi-scale inference to produce accurate regions of interest (ROIs) and confidence scores, reducing computation time by constraining ROI sizes and using multiple models with overlapping size ranges to detect a broader range of object sizes efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-accuracy ML models are used for object detection, then detection accuracy is improved, but computation time increases and real-time performance deteriorates
Solution Approach 1:
The patent segments the object detection task into multiple scales by training separate ML models for different object size ranges. Each model is specialized for a specific scale, allowing faster inference while maintaining accuracy for objects within its size range. This segmentation resolves the contradiction by distributing the computational load across multiple specialized models rather than using one large general model.
Solution Approach 2:
The patent implements dynamic model selection based on the scale of detected objects. The system adaptively chooses which ML model to use depending on the size range of the target object, optimizing computation time by selecting the most appropriate model for each detection task. This dynamic approach maintains high accuracy while minimizing computation time.
2Adaptability or versatility
If ML models are trained to identify all object sizes, then detection coverage is improved, but model complexity and computation requirements increase
Solution Approach 1:
The patent divides the detection coverage requirement into multiple segments, with each ML model handling a specific object size range. This segmentation allows the system to achieve comprehensive detection coverage across all object sizes while keeping each individual model relatively simple and computationally efficient.
Solution Approach 2:
The patent creates a universal detection system that handles all object sizes by combining multiple specialized models. Each model is trained on a specific size range but the overall system achieves universality by selecting the appropriate model based on object scale, maintaining both specialized performance and broad coverage.
3Measurement precision
If ML models process objects outside receptive field at full resolution, then detection accuracy is improved, but memory requirements and computation cost increase
Solution Approach 1:
The patent applies local quality by processing different regions of the image at different resolutions. Objects within the receptive field are processed at full resolution for high accuracy, while objects outside the receptive field are processed at reduced resolution. This local differentiation maintains detection accuracy for critical objects while reducing overall memory and computation requirements.
Data Source
AI summary
Techniques for utilizing multiple scales of images as input to machine learning (ML) models are discussed herein. Operations can include providing an image associated with a first scale to a first ML model. An output of the first ML model can include a first bounding box indicative of a first region of the image representing a first object, with the first bounding box falling within a first range of sizes. Next, a scaled image can be generated by scaling the image. The scaled image can be provided to a second ML model, which can output a second bounding box indicative of a second region of the image representing a second object, the second bounding falling within a second range of sizes. Thus, inputting a scaled image to a same ML model (or to different ML models) can result in different detected features in the images.