Paper für die BMVC 2022 akzeptiert

Das Paper mit dem Titel "Pseudo-Label Noise Suppression Techniques for Semi-Supervised Semantic Segmentation" von Sebastian Scherer, Robin Schön und Rainer Lienhart wurde für die British Machine Vision Conference (BMVC) 2023 akzeptiert. In diesem Paper beschreiben die Autoren eine Methode die es ermöglich den Bedarf an großen gelabelten Datensätzen zu verringern, indem nicht gelabelte Daten in das Training einbezogen werden. Als Anwendung verwenden die Autoren die menschlicher Posenschätzung sowie die semantische Segmentierung, wobei besonderes letzteres interessant ist, da hier die Annotation von Daten äußerst zeitaufwendig ist.


Semi-supervised learning (SSL) can reduce the need for large labelled datasets by incorporating unlabelled data into the training. This is particularly interesting for semantic segmentation, where labelling data is very costly and time-consuming. Current SSL approaches use an initially supervised trained model to generate predictions for unlabelled images, called pseudo-labels, which are subsequently used for training a new model from scratch. Since the predictions usually do not come from an error-free neural network, they are naturally full of errors. However, training with partially incorrect labels often reduce the final model performance. Thus, it is crucial to manage errors/noise of pseudo-labels wisely. In this work, we use three mechanisms to control pseudo-label noise and errors: (1) We construct a solid base framework by mixing images with cow-patterns on unlabelled images to reduce the negative impact of wrong pseudo-labels. Nevertheless, wrong pseudo-labels still have a negative impact on the performance. Therefore, (2) we propose a simple and effective loss weighting scheme for pseudo-labels defined by the feedback of the model trained on these pseudo-labels. This allows us to soft-weight the pseudo-label training examples based on their determined confidence score during training. (3) We also study the common practice to ignore pseudo-labels with low confidence and empirically analyse the influence and effect of pseudo-labels with different confidence ranges on SSL and the contribution of pseudo-label filtering to the achievable performance gains. We show that our method performs superior to state of-the-art alternatives on various datasets. Furthermore, we show that our findings also transfer to other tasks such as human pose estimation.