Automated Model Evaluation for Object Detection via Prediction Consistency and Reliability
This work addresses the need for scalable and cost-effective evaluation of object detection models, which is incremental as it builds on existing AutoEval methods.
The paper tackles the problem of costly manual annotation for evaluating object detectors in real-world applications by developing an automated model evaluation framework (AutoEval) that uses Prediction Consistency and Reliability (PCR) to estimate detection performance without ground-truth labels, achieving more accurate performance estimates than existing methods.
Recent advances in computer vision have made training object detectors more efficient and effective; however, assessing their performance in real-world applications still relies on costly manual annotation. To address this limitation, we develop an automated model evaluation (AutoEval) framework for object detection. We propose Prediction Consistency and Reliability (PCR), which leverages the multiple candidate bounding boxes that conventional detectors generate before non-maximum suppression (NMS). PCR estimates detection performance without ground-truth labels by jointly measuring 1) the spatial consistency between boxes before and after NMS, and 2) the reliability of the retained boxes via the confidence scores of overlapping boxes. For a more realistic and scalable evaluation, we construct a meta-dataset by applying image corruptions of varying severity. Experimental results demonstrate that PCR yields more accurate performance estimates than existing AutoEval methods, and the proposed meta-dataset covers a wider range of detection performance. The code is available at https://github.com/YonseiML/autoeval-det.