Benchmark Analysis of YOLO Performance on Edge Intelligence Devices

Feng, Haogang; Mu, Gaoze; Zhong, Shida; Zhang, Peichang; Yuan, Tao

doi:10.3390/cryptography6020016

Cited by 35 publications

(18 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We opted for YOLOv7 because of its impressive accuracy and superb capability to handle various input images, in addition to its versatility to be deployed on different platforms [ 33 , 34 ]. We have also compared three different deep learning models and confirmed the outstanding performance of YOLO at a reasonably low cost.…”

Section: Discussion and Resultsmentioning

confidence: 99%

Applying a Deep-Learning-Based Keypoint Detection in Analyzing Surface Nanostructures

Yuan

Zhu

et al. 2023

Molecules

View full text Add to dashboard Cite

Scanning tunneling microscopy (STM) imaging has been routinely applied in studying surface nanostructures owing to its capability of acquiring high-resolution molecule-level images of surface nanostructures. However, the image analysis still heavily relies on manual analysis, which is often laborious and lacks uniform criteria. Recently, machine learning has emerged as a powerful tool in material science research for the automatic analysis and processing of image data. In this paper, we propose a method for analyzing molecular STM images using computer vision techniques. We develop a lightweight deep learning framework based on the YOLO algorithm by labeling molecules with its keypoints. Our framework achieves high efficiency while maintaining accuracy, enabling the recognitions of molecules and further statistical analysis. In addition, the usefulness of this model is exemplified by exploring the length of polyphenylene chains fabricated from on-surface synthesis. We foresee that computer vision methods will be frequently used in analyzing image data in the field of surface chemistry.

show abstract

Section: Discussion and Resultsmentioning

confidence: 99%

Applying a Deep-Learning-Based Keypoint Detection in Analyzing Surface Nanostructures

Yuan

Zhu

et al. 2023

Molecules

View full text Add to dashboard Cite

show abstract

“…The second aspect that needs to be considered is when implementing smart applications to an SBCs-based GPU. Because of the memory sharing between CPU and GPU, the architecture and other related parameters must be designed carefully to achieve satisfying accuracy and speed results [16].…”

Section: A the Materialsmentioning

confidence: 99%

Optimizing Hand Gesture Recognition Using CNN Model Supported by Raspberry pi for Self-Service Technology

Rangkuti

Athalaa

Indallah

et al. 2023

JOIV : Int. J. Inform. Visualization

View full text Add to dashboard Cite

This study describes the optimization of hand gesture recognition on Raspberry Pi 4 technology has advanced over the past years, some computers are now able to compute much more complex problems like real-time object detection. But for small devices, optimization is required to run in real-time with acceptable performance in terms of latency and low-cost effect on accuracy. Low latency is a requirement for most technology, especially when integrating real-time object detection as input into Self-Service Technology on Raspberry Pi for the store. This research was conducted on 288 pictures with six types of chosen hand gestures for command inputs that have been configured in the Self-Service Technology as a training dataset. In the experiment carried out with 5 CNN object detection models were used, namely YOLOv3-Tiny-PRN, YOLOv4-Tiny, MobileNetV2-Yolov3-NANO, YOLO-Fastest-1.1, and YOLO-Fastest-1.1-XL. Based on the experiment after optimization, the FPS and inference time metrics have improved performance. The performance improves due to a gained average value of FPS by 3 FPS and a reduced average value of inference time by 119,260 ms. But such an improvement also comes with a reduction in overall accuracy. The rest of the parameters have a reduced score on Precision, Recall, F1-Score, and some for IoU. Only YOLO-Fastest-1.1-XL have an improved value of IoU by about 0.58%. Some improvements in the CNN and dataset might improve the performance even more without sacrificing too much on the accuracy, but it's most likely suitable for another research as a continuation of this topic.

show abstract

“…Additionally, OpenVINO supports accuracy-aware quantization, an advanced method that ensures model accuracy remains within a predefined range by selectively leaving some network layers unquantized. Several authors [36][37][38] leveraged the OpenVINO Toolkit to optimize and accelerate diverse CNN models for various computer vision tasks. However, unfortunately, OpenVINO cannot support pure-integer quantization since it still performs some operations/layers as a floating point.…”

Section: Related Workmentioning

confidence: 99%

Unified Scaling-Based Pure-Integer Quantization for Low-Power Accelerator of Complex CNNs

Al-Hamid

Kim

2023

Electronics

View full text Add to dashboard Cite

Although optimizing deep neural networks is becoming crucial for deploying the networks on edge AI devices, it faces increasing challenges due to scarce hardware resources in modern IoT and mobile devices. This study proposes a quantization method that can quantize all internal computations and parameters in the memory modification. Unlike most previous methods that primarily focused on relatively simple CNN models for image classification, the proposed method, Unified Scaling-Based Pure-Integer Quantization (USPIQ), can handle more complex CNN models for object detection. USPIQ aims to provide a systematic approach to convert all floating-point operations to pure-integer operations in every model layer. It can significantly reduce the computational overhead and make it more suitable for low-power neural network accelerator hardware consisting of pure-integer datapaths and small memory aimed at low-power consumption and small chip size. The proposed method optimally calibrates the scale parameters for each layer using a subset of unlabeled representative images. Furthermore, we introduce a notion of the Unified Scale Factor (USF), which combines the conventional two-step scaling processes (quantization and dequantization) into a single process for each layer. As a result, it improves the inference speed and the accuracy of the resulting quantized model. Our experiment on YOLOv5 models demonstrates that USPIQ can significantly reduce the on-chip memory for parameters and activation data by ~75% and 43.68%, respectively, compared with the floating-point model. These reductions have been achieved with a minimal loss in mAP@0.5—at most 0.61%. In addition, our proposed USPIQ exhibits a significant improvement in the inference speed compared to ONNX Run-Time quantization, achieving a speedup of 1.64 to 2.84 times. We also demonstrate that USPIQ outperforms the previous methods in terms of accuracy and hardware reduction for 8-bit quantization of all YOLOv5 versions.

show abstract

Benchmark Analysis of YOLO Performance on Edge Intelligence Devices

Cited by 35 publications

References 30 publications

Applying a Deep-Learning-Based Keypoint Detection in Analyzing Surface Nanostructures

Applying a Deep-Learning-Based Keypoint Detection in Analyzing Surface Nanostructures

Optimizing Hand Gesture Recognition Using CNN Model Supported by Raspberry pi for Self-Service Technology

Unified Scaling-Based Pure-Integer Quantization for Low-Power Accelerator of Complex CNNs

Contact Info

Product

Resources

About