密集蘑菇簇会严重影响蘑菇质量和自动采摘成功率。为避免形成超密集蘑菇簇,提出一种蘑菇生长状态时空预测算法,对蘑菇生长状态进行预测以指导提前疏蕾。该算法采用编码器-预测器框架,将历史序列图像转换为3D张量序列作为模型的输入;编...密集蘑菇簇会严重影响蘑菇质量和自动采摘成功率。为避免形成超密集蘑菇簇,提出一种蘑菇生长状态时空预测算法,对蘑菇生长状态进行预测以指导提前疏蕾。该算法采用编码器-预测器框架,将历史序列图像转换为3D张量序列作为模型的输入;编码器网络中将卷积和长短时记忆(Long short term memory, LSTM)网络融合实现对蘑菇生长的时空相关性特征的提取;在预测网络中加入扩散模型以解决预测图像的模糊问题;此外,在损失函数中增加了蘑菇面积差异损失函数来进一步减小预测蘑菇与实际蘑菇的形状和位置偏差。实验结果表明,本文算法峰值信噪比可达35.611 dB、多层级结构相似性为0.927、蘑菇预测准确性高达0.93,有效提高了蘑菇生长状态图像预测质量和精度,为食用菌生长预测提供了一种新思路。展开更多
Loss and fragmentation of natural ecosystems are widely recognized as the most important threats to biodiversity conservation, with Neotropical dry forests among the most endangered ecosystems. Area and edge effects a...Loss and fragmentation of natural ecosystems are widely recognized as the most important threats to biodiversity conservation, with Neotropical dry forests among the most endangered ecosystems. Area and edge effects are major factors in fragmented landscapes. Here, we examine area and edge effects and their interaction, on ensembles of arthropods associated to native vegetation in a fragmented Chaco Serrano forest. We analyzed family richness and community composition of herbivores, predators, and parasitoids on three native plant species in 12 fragments of varying size and at edge/interior positions. We also looked for indicator families by using Indicator Species Analysis. Loss of family richness with the reduction of forest fragment area was observed for the three functional groups, with sirnilar magnitude. Herbivores were richer at the edges without interaction between edge and area effects, whereas predators were not affected by edge/interior position and parasitoid richness showed an interaction between area and position, with a steeper area slope at the edges. Family composition of herbivore, predator, and parasitoid assemblages was also affected by forest area and/or edge/interior situation. We found three indicator families for large remnants and five for edges. Our results support the key role of forest area for conservation of arthropods taxonomic and functional diversity in a highly threatened region, and emphasize the need to understand the interactions between area and edge effects on such diversity.展开更多
文摘密集蘑菇簇会严重影响蘑菇质量和自动采摘成功率。为避免形成超密集蘑菇簇,提出一种蘑菇生长状态时空预测算法,对蘑菇生长状态进行预测以指导提前疏蕾。该算法采用编码器-预测器框架,将历史序列图像转换为3D张量序列作为模型的输入;编码器网络中将卷积和长短时记忆(Long short term memory, LSTM)网络融合实现对蘑菇生长的时空相关性特征的提取;在预测网络中加入扩散模型以解决预测图像的模糊问题;此外,在损失函数中增加了蘑菇面积差异损失函数来进一步减小预测蘑菇与实际蘑菇的形状和位置偏差。实验结果表明,本文算法峰值信噪比可达35.611 dB、多层级结构相似性为0.927、蘑菇预测准确性高达0.93,有效提高了蘑菇生长状态图像预测质量和精度,为食用菌生长预测提供了一种新思路。
文摘Loss and fragmentation of natural ecosystems are widely recognized as the most important threats to biodiversity conservation, with Neotropical dry forests among the most endangered ecosystems. Area and edge effects are major factors in fragmented landscapes. Here, we examine area and edge effects and their interaction, on ensembles of arthropods associated to native vegetation in a fragmented Chaco Serrano forest. We analyzed family richness and community composition of herbivores, predators, and parasitoids on three native plant species in 12 fragments of varying size and at edge/interior positions. We also looked for indicator families by using Indicator Species Analysis. Loss of family richness with the reduction of forest fragment area was observed for the three functional groups, with sirnilar magnitude. Herbivores were richer at the edges without interaction between edge and area effects, whereas predators were not affected by edge/interior position and parasitoid richness showed an interaction between area and position, with a steeper area slope at the edges. Family composition of herbivore, predator, and parasitoid assemblages was also affected by forest area and/or edge/interior situation. We found three indicator families for large remnants and five for edges. Our results support the key role of forest area for conservation of arthropods taxonomic and functional diversity in a highly threatened region, and emphasize the need to understand the interactions between area and edge effects on such diversity.
