Multi_layer graph plant leaf segmentation

We showcase the results of our graph-based diffusion technique utilizing random walks with restarts on a multi-layered graph using the publicly accessible Pl@ntleaves (H. Go¨eau, P. Bonnet, A. Joly, N. Boujemaa, D. Barth´el´emy, J.-F. Molino, P. Birnbaum, E. Mouysset, M. Picard, The clef 2011 plant images classification task, Vol. 1177, 2011.) dataset. This dataset comprises 233 high-resolution leaf images taken in their natural environments, presenting various segmentation challenges such as shadows, diverse lighting conditions, and leaf overlap. Our method primarily focuses on identifying leaf regions by initially locating the leaves within the images and then propagating intensity scores from foreground templates to image boundaries to generate saliency maps. By applying a threshold to these saliency maps produced through the diffusion process, we derive binary masks that effectively separate the leaves from the backgrounds. Ground truth images are provided for visual evaluation of our algorithm's performance.Folders description: image: RGB images mask: Ground truth masks FG_templates: foreground templates and bounding boxes defined on dataset images
Salinecy_map: saliency maps obtained by our approach PR_masks: Predicted masks obtained by tresholding our salinecy maps Plant_Leaf_Segmentation: a compressed folder containing the above folders.

本项研究展示了我们基于图扩散技术的成果，该技术采用随机游走重启策略在多层图上进行，并使用公开可得的Pl@ntleaves数据集（H. Go¨eau, P. Bonnet, A. Joly, N. Boujemaa, D. Barth´el´emy, J.-F. Molino, P. Birnbaum, E. Mouysset, M. Picard, The clef 2011 plant images classification task, Vol. 1177, 2011）进行验证。该数据集包含233张在自然环境中拍摄的高分辨率叶片图像，呈现了诸如阴影、多变的照明条件和叶片重叠等多样化的分割挑战。本研究方法主要致力于通过首先在图像中定位叶片，然后将前景模板的强度分数传播至图像边界以生成显著性图来识别叶片区域。通过对扩散过程产生的显著性图应用阈值，我们得到了能够有效将叶片与背景分离的二值掩码。同时提供了地面真实图像以供对算法性能进行视觉评估。文件夹描述如下：

image: RGB图像
mask: 地面真实掩码
FG_templates: 数据集图像上定义的前景模板和边界框
Salience_map: 通过我们的方法获得的显著性图
PR_masks: 通过对显著性图进行阈值处理得到的预测掩码
Plant_Leaf_Segmentation: 包含上述文件夹的压缩文件夹。