ML Object Detection via Image Segmentation for Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection systems struggle when processing images containing multiple instances of the same type of object, often resulting in inaccurate results.
Innovation Solution
The system employs a machine learning model to detect multiple objects in an image, generates separate images for each object, converts text to machine-readable format, and uses additional machine learning models to determine attribute values, thereby processing each object as a separate image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional object detection systems process images containing multiple instances of the same type of object, then processing efficiency is maintained, but detection accuracy deteriorates
Solution Approach 1:
The system segments an image containing multiple objects of the same type into multiple separate images, each containing a single object instance. This segmentation is achieved by detecting each object instance and extracting it as an independent image. By processing each segmented image separately through the object detection system, the accuracy of detection for each individual object is significantly improved while maintaining overall processing efficiency through automated batch handling.
2Reliability
If a single machine learning model processes all objects in an image, then device complexity is minimized, but detection reliability deteriorates
Solution Approach 1:
The system divides the detection task into multiple independent processing streams by segmenting the original image into separate images for each object instance. Each segmented image is then processed independently, allowing the system to maintain simpler individual processing pipelines while achieving higher overall reliability through the combination of multiple independent detections.
Solution Approach 2:
The system creates multiple copies of the detection processing pipeline, with each copy handling a separate segmented image. This copying approach allows the same detection algorithm to be applied independently to each object instance, improving reliability by eliminating cross-contamination between object detections while maintaining consistent processing logic across all instances.
Data Source
AI summary
Some embodiments provide a non-transitory machine-readable medium that stores a program executable by a device. The program receives a request to process an image for multiple objects. The program further uses a machine learning model to detect a plurality of objects in the image. The program also generates a plurality of images based on the plurality of objects in the image. For each image in the plurality of images, the program further converts text in the image to machine-readable text. For each image in the plurality of images, the program also uses a set of machine learning models to determine a set of values for a set of attributes. For each set of values determined for the set of attributes, the program further generates a record comprising the set of attributes and storing the set of values for the set of attributes in the record.