Markerless Dog Pose Recognition in the Wild Using ResNet Deep Learning Model

Srinivasan, Raman; Maskeliūnas, Rytis; Damaševičius, Robertas

doi:10.3390/computers11010002

Cited by 10 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the reasons is that there are still many limitations in the real-time performance of these solutions. For example, the hardware of OpenPose that can achieve real-time is limited to the current high-end GPUs, and it cannot achieve the real-time requirement on the low-end GPUs [14]. Many 3D pose estimation techniques that claim to be real-time rely on high-quality 2D pose input, which takes time to obtain.…”

Section: Introductionmentioning

confidence: 99%

Convolution-Based Design for Real-Time Pose Recognition and Character Animation Generation

Wang

Lee

2022

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

Human pose recognition and its generation are an important animation design key point. To this end, this paper designs new neural network structures for 2D and 3D pose extraction tasks and corresponding GPU-oriented acceleration schemes. The scheme first takes an image as input, extracts the human pose from it, converts it into an abstract pose data structure, and then uses the converted dataset as a basis to generate the desired character animation based on the input at runtime. The scheme in this paper has been tested on pose recognition datasets and different levels of hardware showing that 2D pose recognition can reach speeds above 60 fps on common computer hardware, 3D pose recognition can be estimated to reach speeds above 24 fps with an average error of only 110 mm, and real-time animation generation can reach speeds above 30 frames per second.

show abstract

Section: Introductionmentioning

confidence: 99%

Convolution-Based Design for Real-Time Pose Recognition and Character Animation Generation

Wang

Lee

2022

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

show abstract

“…Unlike traditional CNN, ResNet is composed of multiple residual blocks in series, which can easily adjust the width and depth of the network to obtain networks with different expression capabilities and does not need to worry about the problem of gradient disappearance (Ning et al, 2020b ). It has more powerful adaptive feature extraction capabilities for two-dimensional medical image data (Raman et al, 2021 ; Ahamed et al, 2023 ). The residual structure is shown in Figure 5 .…”

Section: Methodsmentioning

confidence: 99%

Chaotic medical image encryption method using attention mechanism fusion ResNet model

Peng

2023

Front. Neurosci.

View full text Add to dashboard Cite

IntroductionWith the rapid advancement of artificial intelligence (AI) technology, the protection of patient medical image privacy and security has become a critical concern in current research on image privacy protection. However, traditional methods for encrypting medical images have faced criticism due to their limited flexibility and inadequate security. To overcome these limitations, this study proposes a novel chaotic medical image encryption method, called AT-ResNet-CM, which incorporates the attention mechanism fused with the ResNet model.MethodsThe proposed method utilizes the ResNet model as the underlying network for constructing the encryption and decryption framework. The ResNet's residual structure and jump connections are employed to effectively extract profound information from medical images and expedite the model's convergence. To enhance security, the output of the ResNet model is encrypted using a logistic chaotic system, introducing randomness and complexity to the encryption process. Additionally, an attention mechanism is introduced to enhance the model's response to the region of interest within the medical image, thereby strengthening the security of the encrypted network.ResultsExperimental simulations and analyses were conducted to evaluate the performance of the proposed approach. The results demonstrate that the proposed method outperforms alternative models in terms of encryption effectiveness, as indicated by a horizontal correlation coefficient of 0.0021 and information entropy of 0.9887. Furthermore, the incorporation of the attention mechanism significantly improves the encryption performance, reducing the horizontal correlation coefficient to 0.0010 and increasing the information entropy to 0.9965. These findings validate the efficacy of the proposed method for medical image encryption tasks, as it offers enhanced security and flexibility compared to existing approaches.DiscussionIn conclusion, the AT-ResNet-CM method presents a promising solution to address the limitations of traditional encryption techniques in protecting patient medical images. By leveraging the attention mechanism fused with the ResNet model, the method achieves improved security and flexibility. The experimental results substantiate the superiority of the proposed method in terms of encryption effectiveness, horizontal correlation coefficient, and information entropy. The proposed method not only addresses the shortcomings of traditional methods but also provides a more robust and reliable approach for safeguarding patient medical image privacy and security.

show abstract

“…Maskeliunas et al [11] analyzed emotions of dogs from their vocalizations, using cochleagrams to categorize dog barking into the categories angry, crying, happy, and lonely. Raman et al [12] used CNNs and DeepLabCut to automatically recognize dog body parts and dog poses.…”

Section: Introductionmentioning

confidence: 99%

Predicting Dog Emotions Based on Posture Analysis Using DeepLabCut

2022

View full text Add to dashboard Cite

This paper describes an emotion recognition system for dogs automatically identifying the emotions anger, fear, happiness, and relaxation. It is based on a previously trained machine learning model, which uses automatic pose estimation to differentiate emotional states of canines. Towards that goal, we have compiled a picture library with full body dog pictures featuring 400 images with 100 samples each for the states “Anger”, “Fear”, “Happiness” and “Relaxation”. A new dog keypoint detection model was built using the framework DeepLabCut for animal keypoint detector training. The newly trained detector learned from a total of 13,809 annotated dog images and possesses the capability to estimate the coordinates of 24 different dog body part keypoints. Our application is able to determine a dog’s emotional state visually with an accuracy between 60% and 70%, exceeding human capability to recognize dog emotions.

show abstract

Markerless Dog Pose Recognition in the Wild Using ResNet Deep Learning Model

Cited by 10 publications

References 37 publications

Convolution-Based Design for Real-Time Pose Recognition and Character Animation Generation

Convolution-Based Design for Real-Time Pose Recognition and Character Animation Generation

Chaotic medical image encryption method using attention mechanism fusion ResNet model

Predicting Dog Emotions Based on Posture Analysis Using DeepLabCut

Contact Info

Product

Resources

About