Non-destructive and online defect detection on seals is increasingly being deployed inpackaging processes, especially for food and pharmaceutical products. It is a key controlstep in these processes as it curtails the costs of these defects.To address this cause, this paper highlights a combination of two cost-effective methods,namely machine learning algorithms and infrared thermography. Expectations can,however, be restricted when the training data is small, unbalanced, and subject to

more »

... mperfections.This paper proposes a classification method that tackles these limitations. Its accuracyexceeds 93% with two small training sets, including 2.5 to 10 times fewer negatives. Itsalgorithm has a low computational cost compared to deep learning approaches, and doesnot need any prior statistical studies on defects characterization.