Detecção E Reconhecimento De Plantas De Pequeno Porte Utilizando Aprendizagem De Máquina

Abstract

The detection and recognition of plants has always been a difficult task even for connoisseurs and scholars due to the wide variety of plants found worldwide. With the advancement of technology, it has become possible to solve this problem computationally. This paper presents a method to perform plant detection and recognition from images using computer vision and artificial intelligence algorithms. The results show that the computational cost and recognition rate were satisfactory for use in controlled environments. The processing time to recognize each plant was 375 milliseconds, with an accuracy of 92%.

Deep learning based slope erosion detection

“…In practice, the outputs of the filters are submitted to the activation function at the end of each and every convolutional layer, and after going through this process, used to update the neural network weights [39]. By observing the mathematical expression of the ReLU function, 𝑓(𝑥) = max(0, 𝑥), it can be seen that the neurons will only activate if the input is greater than zero, so, neurons that receive negative values will be "erased".…”