Bird posture recognition based on target keypoints estimation in dual-task convolutional neural networks

Lin, Chia-Hsien; Hong, Sidi; Lin, Mengxiang; Huang, Xiuping; Liu, Jinfu

doi:10.1016/j.ecolind.2021.108506

Cited by 16 publications

(11 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although a key advantage of using a motion-capture system is its ability to track 3D coordinates of reflective markers in high tracking accuracy, key shortcomings are the necessity of attaching small markers to the birds and also the time and cost needed to establish the motion-capture system. An alternative method is using multi-array cameras and deep-learning system to achieve markerless tracking of birds' postures 39,40 (also see [41][42][43] ). Although the markerless tracking technique should be applicable in a wider variety of study situations as compared to the motion-capture technique, it is currently not compelling with the motion-capture technique in terms of tracking accuracy and robustness.…”

Section: Discussionmentioning

confidence: 99%

“…However, this method is not able to analyze the 6 degrees-of-freedom (DOF) movements of birds' heads. A potential, emerging alternative technique is using multi-array camera images and deep learning algorithms to estimate 3-dimensional (3D) body postures including 6 DOF head movements 39,40 (also see [41][42][43] ), but the accuracy and applicability of this approach have not yet been thoroughly explored in birds. Another technique, and the one that we will adopt here, is to use a motion-capture system that can record the 3D coordinates of small reflective markers attached to a bird's head with high spatiotemporal resolutions.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Head-tracking of freely-behaving pigeons in a motion-capture system reveals the selective use of visual field regions

Kano

Naik

Keskin

et al. 2022

Sci Rep

View full text Add to dashboard Cite

Using a motion-capture system and custom head-calibration methods, we reconstructed the head-centric view of freely behaving pigeons and examined how they orient their head when presented with various types of attention-getting objects at various relative locations. Pigeons predominantly employed their retinal specializations to view a visual target, namely their foveas projecting laterally (at an azimuth of ± 75°) into the horizon, and their visually-sensitive “red areas” projecting broadly into the lower-frontal visual field. Pigeons used their foveas to view any distant object while they used their red areas to view a nearby object on the ground (< 50 cm). Pigeons “fixated” a visual target with their foveas; the intervals between head-saccades were longer when the visual target was viewed by birds’ foveas compared to when it was viewed by any other region. Furthermore, pigeons showed a weak preference to use their right eye to examine small objects distinctive in detailed features and their left eye to view threat-related or social stimuli. Despite the known difficulty in identifying where a bird is attending, we show that it is possible to estimate the visual attention of freely-behaving birds by tracking the projections of their retinal specializations in their visual field with cutting-edge methods.

show abstract

Section: Discussionmentioning

confidence: 99%

mentioning

confidence: 99%

Head-tracking of freely-behaving pigeons in a motion-capture system reveals the selective use of visual field regions

Kano

Naik

Keskin

et al. 2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…With the development of artificial intelligence technology, scholars have used artificial intelligence methods to automatically locate acupuncture points [4][5][6][7], and acupuncture points are located through traditional algorithms such as ASM, ARM, and PSO, but they exist problems such as difficult application, complex learning process, long training time and low positioning accuracy. In recent years, deep learning has been applied to many fields, among which the use of deep learning for target recognition and key point positioning has been a hot topic [8][9][10], due to RCNN [11], SPPNet [12] and YOLO [13] and other target recognition and keypoint positioning networks are proposed, and the automatic positioning of acupuncture points has reached a new era of development. Based on the deep learning model and combined with the fingering method, Wang Cong made a preliminary positioning of the hand acupuncture points, and then made a depth estimation of the acupuncture point positions, and obtained more accurate hand acupuncture point coordinates [14].…”

Section: Introductionmentioning

confidence: 99%

Back acupoint location method based on Deep Learning

Liu

Qin

Zeng

2022

Preprint

View full text Add to dashboard Cite

Aiming at the problem of automatic and accurate positioning of back acupuncture points and evaluation of the positioning accuracy of massage robots or automatic cupping machines, the human back data set is expanded through Deep Convolution Generative Adversarial Networks (DCGAN), marking the back acupoints with the method of bone grading, and then use the Keypoint region-based convolutional neural network (Keypoint RCNN) to train the back acupoint location model. Because the size and area of the acupuncture points are difficult to measure, there is no clear Based on the evaluation index of acupoint positioning, and finally combined with the method proposed in the literature and the experience of back acupoint searching, a back acupoint positioning evaluation method was established. According to this method, the average accuracy rate of the back acupoint positioning model reached 86.33%, which can locate the back acupoints automatically and accurately, therefore it is beneficial to the application of back acupoints in treatment and health care.

show abstract

“…However, this method is not able to analyze the 6 degrees-offreedom (DOF) movements of birds' heads. A potential, emerging alternative technique is using multiarray camera images and deep learning algorithms to estimate 3-dimensional (3D) body postures including 6 DOF head movements 39 (also see 40 ), but the accuracy and applicability of this approach have not yet been thoroughly explored in birds. Another technique, and the one that we will adopt here, is to use a well-calibrated motion-capture system that can record the 3D coordinates of small re ective markers attached to a bird's head at a high temporal frequency and with submillimeter precision.…”

Section: Introductionmentioning

confidence: 99%

What are birds looking at? A large-scale motion-capture system reveals the visual attention of freely- behaving pigeons

Kano

Naik

Keskin

et al. 2022

Preprint

View full text Add to dashboard Cite

Using a large-scale motion-capture system and custom head-calibration methods, we reconstructed the head-centric view of freely behaving pigeons and examined how they orient their head when presented with various types of attention-getting objects at various relative locations. Pigeons predominantly employed their retinal specializations to view a visual target, namely their foveas projecting laterally (at an azimuth of ±75°) into the horizon, and their visually-sensitive “red areas” projecting broadly into the lower-frontal visual field. Pigeons used their foveas to view any distant object while they used their red areas to view a nearby food-related object on the ground (< 50 cm). Pigeons “fixated” a visual target with their foveas; the intervals between head saccades were longer when the visual target was viewed by birds’ foveas compared to when it was viewed by any other visual field region. Furthermore, pigeons used their right eye to observe food-related objects and their left eye to do so for threat-related or social stimuli. Despite the known difficulty in identifying where a bird is attending, we show that it is possible to estimate the visual attention of freely behaving individuals by tracking the projections of their retinal specializations in their visual field with cutting-edge methods.

show abstract

Bird posture recognition based on target keypoints estimation in dual-task convolutional neural networks

Cited by 16 publications

References 25 publications

Head-tracking of freely-behaving pigeons in a motion-capture system reveals the selective use of visual field regions

Head-tracking of freely-behaving pigeons in a motion-capture system reveals the selective use of visual field regions

Back acupoint location method based on Deep Learning

What are birds looking at? A large-scale motion-capture system reveals the visual attention of freely- behaving pigeons

Contact Info

Product

Resources

About