Application of Deep-Learning Methods to Bird Detection Using Unmanned Aerial Vehicle Imagery

Abstract

Wild birds are monitored with the important objectives of identifying their habitats and estimating the size of their populations. Especially in the case of migratory bird, they are significantly recorded during specific periods of time to forecast any possible spread of animal disease such as avian influenza. This study led to the construction of deep-learning-based object-detection models with the aid of aerial photographs collected by an unmanned aerial vehicle (UAV). The dataset containing the aerial photographs includes diverse images of birds in various bird habitats and in the vicinity of lakes and on farmland. In addition, aerial images of bird decoys are captured to achieve various bird patterns and more accurate bird information. Bird detection models such as Faster Region-based Convolutional Neural Network (R-CNN), Region-based Fully Convolutional Network (R-FCN), Single Shot MultiBox Detector (SSD), Retinanet, and You Only Look Once (YOLO) were created and the performance of all models was estimated by comparing their computing speed and average precision. The test results show Faster R-CNN to be the most accurate and YOLO to be the fastest among the models. The combined results demonstrate that the use of deep-learning-based detection methods in combination with UAV aerial imagery is fairly suitable for bird detection in various environments.

Jellytoring: Real-Time Jellyfish Monitoring Based on Deep Learning Object Detection

“…Over the last decade, automatic detection methods have arisen as a cost-effective way for image location and classification [11], this is highly relevant in regards to the increasing amount of image data that is being collected from the marine environment. In general, images of animal species are used to record and quantify their density, distribution and behavior [12][13][14][15]. Getting to determine where objects are located in a given image (object localization) and which category each object belongs to (object classification) can be useful in a multitude of scenarios and implemented for multiple applications.…”

Learning of Counting Crowded Birds of Various Scales via Novel Density Activation Maps

“…Meanwhile, deep learning methods were developed to detect birds automatically. These studies [2], [8], [19], [21], [25] mostly used variations of RCNN [6] to detect birds by surrounding each bird with a bounding box. The number of birds can be obtained by counting the generated bounding boxes.…”

“…Meanwhile, deep learning methods were applied to the detection of birds where they mostly adopt the RCNN [6] to detect birds in various circumstances with a bounding box surrounding each bird individually [2], [8], [21], [24], [25], [29]. However, their works are limited to situations where the birds are very sparse since the accuracy of bounding box detection methods decreases sharply when the objects get crowded.…”

Robust Object Detection in Aerial Imagery Based on Multi-Scale Detector and Soft Densely Connected

“…[28] makes an elaboration on the potential and limitations of Faster R-CNN when it is used for tracking medium-sized objects in cases of pedestrian detection. Those methods of one stage, such as You Only Look Once (YOLO) [29], Single Shot Multi-Box Detector (SSD) [30] and Retinanet [31], carry out the bounding box and classification processes simultaneously [32]. These two kinds of methods, however, have some differences in their performance regarding to the computing speed and detection accuracy, factors that can also be influenced by the type of CNN backbone employed, such as Googlenet [33], VGGNet [12], Resnet [13], Darknet-53 [29] or Densenet [34].…”