Insect Identification Among Deep Learning’s Meta-architectures Using TensorFlow

Abstract

Agriculture provides food for human existence, where insects damage the crops. The identification of the insect is a difficult process and subjected to expert opinion. In recent years, researches using deep learning in fields of object detection have been widespread and show accuracy as a result. This study show the comparison of three widely used deep learning meta-architectures (Faster R-CNN, SSD Inception and SSD Mobilenet) as object detection for selected flying insects namely Phyllophaga spp., Helicoverpa armigera and Spodoptera litura. The proposed study is focused on accuracy performance of selected meta-architectures using small dataset of insects. The meta-architecture was tested with same environment for all three architectures and Faster RCNN meta-architecture performs outstanding with an accuracy of 95.33%

A Systematic Literature Review on Insect Detection in Images

“…In the studied articles, only 11 of them mentioned the tool used to label the images. Nam et al [35] mentioned that they used BBOX [60] while other authors [10], [14], [17], [25], [28], [33], [44] used LabelImg [61] and further authors [27], [30], [50] used LabelMe [62]. Some articles [35], [42], [44] mention that the labelling is done by experts, and Rustia et al [33] state that labelling was done by experts and entomologists, while Ding et al [19] [40] did not specify which method they employed.…”

“…In some cases, it is used alone [25], [41], [89] while Nam et al [35] use SSD with VGG16. Finally, Patel et al [28] used a combination of three techniques: Faster R-CNN and SSD with two different feature extractors: Inception and MobileNet.…”

“…There was an insect species called the moth that was present in numerous studies, and it has been a serious problem for farmers around the world. This insect has been detected in several publications [19], [23], [28], [30], [41], [50], [52]. whereas Patel et al [28] detect three insects where two of them are moths, and Korsch et al [41] detect this type of insect in 200 species.…”

“…This insect has been detected in several publications [19], [23], [28], [30], [41], [50], [52]. whereas Patel et al [28] detect three insects where two of them are moths, and Korsch et al [41] detect this type of insect in 200 species.…”

“…-Some articles only detect one insect or one class: [9], [19], [23], [30], [41], [42], [44], [50], [52], [82], [90], -Others detect two classes: [8], [14], [26], [46], -Other papers detect 3 classes: [7], [20], [27], [28], [32], [48], [51], -Other articles detect 4 classes: [17], and [89], -Some of the studied papers detect 5 classes: [35], [10], and [39], -There are some papers that detect 6 classes: [15], [16], [18], [25], [34], -Other papers detect from 7 to 9 classes: [24] (7 classes), [31], [37], and [85] Since there are fewer studies as the number of insects increases, 66% of all articles focus on insects with a population of one to six.…”