ePillID Dataset: A Low-Shot Fine-Grained Benchmark for Pill Identification
This work tackles the problem of medication errors for patients and medical professionals by providing a low-shot fine-grained benchmark, though it is incremental as it focuses on dataset creation and baseline evaluation.
The authors introduced ePillID, a large public benchmark for pill image recognition with 13k images across 9804 appearance classes, addressing the error-prone task of identifying prescription medications. The best baseline model using a multi-head metric-learning approach performed well but still failed to distinguish particularly confusing classes.
Identifying prescription medications is a frequent task for patients and medical professionals; however, this is an error-prone task as many pills have similar appearances (e.g. white round pills), which increases the risk of medication errors. In this paper, we introduce ePillID, the largest public benchmark on pill image recognition, composed of 13k images representing 9804 appearance classes (two sides for 4902 pill types). For most of the appearance classes, there exists only one reference image, making it a challenging low-shot recognition setting. We present our experimental setup and evaluation results of various baseline models on the benchmark. The best baseline using a multi-head metric-learning approach with bilinear features performed remarkably well; however, our error analysis suggests that they still fail to distinguish particularly confusing classes. The code and data are available at https://github.com/usuyama/ePillID-benchmark.