Object Detection Device Using Temporal Similarity for Reliable Target Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In object detection processing, a rectangular frame with a target object may have a confidence score below the threshold due to variations in the object's state, leading to its deletion and failure in detection.
Innovation Solution
An object detection device calculates a similarity score between a bounding box from a previous image and candidate rectangular frames in a new image, increasing the confidence score of frames with high similarity and deleting those below the threshold, ensuring accurate detection of target objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deletion processing is performed to remove rectangular frames with confidence scores below the threshold, then false detections are reduced, but valid target objects may be missed when confidence scores decrease due to object state variations
Solution Approach 1:
The system performs preliminary action by using the bounding box from the comparison image (previous frame) to guide the detection in the detection target image (current frame). The calculation unit calculates a similarity score between the comparison bounding box and candidate rectangular frames before deletion processing, allowing frames with lower confidence scores to be retained if they show high similarity to the previous frame's detection, thus preventing false deletions while maintaining reliability
Solution Approach 2:
The system implements feedback by using the detection result from the comparison image to influence the deletion decision in the detection target image. The similarity score calculation provides feedback about the consistency of object detection across frames, allowing the system to adjust which frames to delete based on temporal consistency rather than solely on absolute confidence scores
2Reliability
If the confidence score threshold is set high to ensure detection quality, then false positives are minimized, but detection recall decreases when object appearance varies
Solution Approach 1:
The system performs preliminary action by establishing a high confidence threshold for quality assurance, then uses the similarity score calculation as a preliminary filter to identify candidate frames that should be retained despite lower confidence scores. This allows the system to maintain high detection quality while improving recall by preventing premature elimination of potentially valid detections
Solution Approach 2:
The system applies local quality by differentiating the evaluation criteria for different frames. Instead of applying a uniform threshold to all frames, the system evaluates each frame's similarity score relative to the comparison image and adjusts the deletion decision accordingly. Frames with high similarity scores are retained even with lower confidence scores, while frames with low similarity scores are deleted regardless of confidence score
Data Source
AI summary
An object detection device, which performs object detection processing to detect an area where a target object is present in images captured by a camera, includes a memory storing a bounding box indicating the area where the target object is present, a processor executing program codes or commands, an acquisition unit acquiring a detection target image, a providing unit providing a plurality of rectangular frames in the detection target image, a calculation unit performing calculation processing to calculate a similarity score between the bounding box and the rectangular frame, an addition unit performing addition processing to increase a confidence score, a deletion unit performing deletion processing to delete a rectangular frame having the confidence score less than a confidence score threshold, and an output unit outputting a rectangular frame remaining after the deletion processing as the bounding box.