Automatic detection of floating instream large wood in videos using deep learning

Instream large wood (i.e. downed trees, branches, and roots larger than 1m in length and 10 cm in diameter) performs essential geomorphological and ecological functions that support the health of river ecosystems. However, even though its transport during floods may pose risks, it is rarely observed and remains poorly understood. This paper presents a novel approach for detecting floating pieces of instream wood in videos. The approach uses a convolutional neural network to automatically detect wood. We sampled data to represent different wood transport conditions, combining 20 datasets to yield thousands of instream wood images. We designed multiple scenarios using different data subsets with and without data augmentation. We analysed the contribution of each scenario to the effectiveness of the model using k-fold cross-validation. The mean average precision of the model varies between 35 % and 93 % and is influenced by the quality of the data that the model detects. When using a 418-pixel input image resolution, the model detects wood with an overall mean average precision of 67 %. Improvements in mean average precision of up to 23 % could be achieved in some instances, and increasing the input resolution raised the weighted mean average precision to 74 %. We demonstrate that detection performance on a specific dataset is not solely determined by the complexity of the network or the training data. Therefore, the findings of this paper could be used when designing a custom wood detection network. With the growing availability of flood-related videos featuring wood uploaded to the internet, this methodology facilitates the quantification of wood transport across a wide variety of data sources.