A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Garbage pickup trolley based on residual semantic enhanced deep neural network
The invention provides a garbage pickup trolley based on a residual semantic enhanced deep neural network, which comprises a mechanical structure, a control module, a visual identification and wireless communication module and a mode selection and display module, and integrates a real-time visual garbage classification device and an automatic garbage collection device on one device. A foundation is laid for future development of the smart home Internet of Things; the structure is simple, mass production is easy, equipment cost is low, and wide application is facilitated; a residual semantic enhanced deep neural network is adopted, and the network performs reasoning on an advanced visual concept, so that compared with other models, the network has a relatively good effect and very high garbage classification accuracy, and correct classification of garbage is realized; and meanwhile, three different modes are provided, manpower input can be reduced, the safety of workers is guaranteed on some special occasions, and automatic grabbing and classification of the garbage are achieved.
本发明提供了一种基于残差语义强化深度神经网络的垃圾拾取小车，包括机械结构、控制模块、视觉识别及无线通信模块、模式选择与显示模块，将实时视觉垃圾分类与垃圾自动收集装置集成在一个设备上，为将来的智能家居物联网发展奠定基础；结构简单易于量产，设备成本低益于广泛投入使用；采用了残差语义强化深度神经网络，该网络通过对高级视觉概念进行推理，相对于其他模型有较好的效果，有很高的垃圾分类准确率，实现了垃圾的正确分类；同时本发明提供了三种不同的各种方式，能够减少人力投入，在一些特殊场合保证工作人员的安全，实现垃圾的自动抓取与分类。
Garbage pickup trolley based on residual semantic enhanced deep neural network
The invention provides a garbage pickup trolley based on a residual semantic enhanced deep neural network, which comprises a mechanical structure, a control module, a visual identification and wireless communication module and a mode selection and display module, and integrates a real-time visual garbage classification device and an automatic garbage collection device on one device. A foundation is laid for future development of the smart home Internet of Things; the structure is simple, mass production is easy, equipment cost is low, and wide application is facilitated; a residual semantic enhanced deep neural network is adopted, and the network performs reasoning on an advanced visual concept, so that compared with other models, the network has a relatively good effect and very high garbage classification accuracy, and correct classification of garbage is realized; and meanwhile, three different modes are provided, manpower input can be reduced, the safety of workers is guaranteed on some special occasions, and automatic grabbing and classification of the garbage are achieved.
本发明提供了一种基于残差语义强化深度神经网络的垃圾拾取小车，包括机械结构、控制模块、视觉识别及无线通信模块、模式选择与显示模块，将实时视觉垃圾分类与垃圾自动收集装置集成在一个设备上，为将来的智能家居物联网发展奠定基础；结构简单易于量产，设备成本低益于广泛投入使用；采用了残差语义强化深度神经网络，该网络通过对高级视觉概念进行推理，相对于其他模型有较好的效果，有很高的垃圾分类准确率，实现了垃圾的正确分类；同时本发明提供了三种不同的各种方式，能够减少人力投入，在一些特殊场合保证工作人员的安全，实现垃圾的自动抓取与分类。
Garbage pickup trolley based on residual semantic enhanced deep neural network
一种基于残差语义强化深度神经网络的垃圾拾取小车
SU WEN (author) / XU XINLIN (author) / LI CHUN (author) / HU YUCHAO (author) / HUANG BOHAN (author) / ZHOU PEITING (author) / SU YUAN (author)
2024-10-18
Patent
Electronic Resource
Chinese
Garbage classification and pickup robot based on deep learning
| European Patent Office | 2020