Dense Multi-Scale Convolutional Network for Plant Segmentation

Plant segmentation is a critical task in precision agriculture as related to crop management and weed treatment. Plants can exhibit very large scale changes, which presents great challenge for accurate crop/weed segmentation. Recent works have shown that multi-scale features are useful segment objects with different scales. In this work, we propose Dense Multi-scale Convolutional Network (DMSCN) pixel-wise Our network has an encoder-decoder structure. The encoder comprises of (DCN) Multi-Scale Atrous Pooling (DMSAP) module. DCN composed standard atrous convolutions dense connections. architecture allows the increase density feature maps while avoiding signal decimation due dimension reduction. proposed DMSAP connects set convolutional layers dilation rates densely cascaded manner. able capture sampling receptive field. A simple yet effective decoder used refine results by combining high low-level encoder. Extensive experiments performed on four datasets. One these datasets was collected annotated us. We conduct ablation study show advantages modules DMSCN. comparative demonstrates our model compared previous methods terms accuracy complexity.