论文标题
一种基于轻巧的变压器的鱼层检测模型
A lightweight Transformer-based model for fish landmark detection
论文作者
论文摘要
当有足够的训练数据时，基于变压器的模型（例如Vision Transformer（VIT））可以在某些视力任务中胜过跨倾性神经网络（CNN）。但是，（CNN）对视力任务（即翻译均衡和局部性）具有强大而有用的归纳偏见。在这项工作中，我们开发了一种新颖的模型架构，我们称之为移动鱼类地标检测网络（MFLD-NET）。我们已经使用基于VIT的卷积操作（即斑块嵌入，多层感知器）制作了该模型。 MFLD-NET可以在轻巧的同时获得竞争性或更好的结果，同时轻巧，因此适用于嵌入式和移动设备。此外，我们表明MFLD-NET可以在PAR上获得关键点（地标）估计精度，甚至比FISH Image DataSet上的一些最先进的（CNN）更好。此外，与VIT不同，MFLD-NET不需要预训练的模型，并且在小型数据集中训练时可以很好地概括。我们提供定量和定性的结果，以证明该模型的概括能力。这项工作将为未来开发移动但高效的鱼类监测系统和设备的努力提供基础。
Transformer-based models, such as the Vision Transformer (ViT), can outperform onvolutional Neural Networks (CNNs) in some vision tasks when there is sufficient training data. However, (CNNs) have a strong and useful inductive bias for vision tasks (i.e. translation equivariance and locality). In this work, we developed a novel model architecture that we call a Mobile fish landmark detection network (MFLD-net). We have made this model using convolution operations based on ViT (i.e. Patch embeddings, Multi-Layer Perceptrons). MFLD-net can achieve competitive or better results in low data regimes while being lightweight and therefore suitable for embedded and mobile devices. Furthermore, we show that MFLD-net can achieve keypoint (landmark) estimation accuracies on-par or even better than some of the state-of-the-art (CNNs) on a fish image dataset. Additionally, unlike ViT, MFLD-net does not need a pre-trained model and can generalise well when trained on a small dataset. We provide quantitative and qualitative results that demonstrate the model's generalisation capabilities. This work will provide a foundation for future efforts in developing mobile, but efficient fish monitoring systems and devices.
