Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications
Image processing is an essential first step towards fully utilizing robotics, deep learning, and machine learning techniques. Using techniques like image enhancement, restoration, and segmentation are able to extract pertinent information from images and use it for task execution and decision-making...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-12-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123024014348 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850054773120696320 |
|---|---|
| author | J. Jean Jenifer Nesam S. Sankar Ganesh Sitharthan Ramachandran |
| author_facet | J. Jean Jenifer Nesam S. Sankar Ganesh Sitharthan Ramachandran |
| author_sort | J. Jean Jenifer Nesam |
| collection | DOAJ |
| description | Image processing is an essential first step towards fully utilizing robotics, deep learning, and machine learning techniques. Using techniques like image enhancement, restoration, and segmentation are able to extract pertinent information from images and use it for task execution and decision-making. However, the hardware implementation of these algorithms demands more delay, area, and power. This work proposes a new mantissa bit-size reduced half-precision floating-point format for processing and characterizing image pixels for machine learning algorithms. In the realm of imaging, lowering the mantissa bit-size in floating-point can conserve area and power when utilized for internal calculations. Together with the area-power reduction, there is also a progressive reduction in image quality. For any image application, it becomes necessary to monitor the area and power trade-offs related to the amount of bits used to process the raw data. This work assists in selecting the bit-size for internal computations based on the accuracy requirements of the application by reporting the area, and power for various bit size reductions. The updated pixel values after applying the mantissa bit-size reduction are displayed in this study, along with a theoretical explanation of its inaccuracy. Since multipliers and adders are needed for the majority of mathematical equations in machine learning image algorithms, they are developed later in this work to process the image. The processed image is based on various adjusted pixel values, and the experimental findings demonstrate 75.2% to 21.3% area optimization, and 66.43% to 20.4% power optimization. However, determining the PSNR and MSE values of the processed image allows for quality validation. |
| format | Article |
| id | doaj-art-9855e0375d914a28a607db2a08172cba |
| institution | DOAJ |
| issn | 2590-1230 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj-art-9855e0375d914a28a607db2a08172cba2025-08-20T02:52:09ZengElsevierResults in Engineering2590-12302024-12-012410317910.1016/j.rineng.2024.103179Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applicationsJ. Jean Jenifer Nesam0S. Sankar Ganesh1Sitharthan Ramachandran2School of Electronics Engineering, Vellore Institute of Technology (VIT), Chennai, Tamil Nadu, India; Corresponding author.School of Electronics Engineering, Vellore Institute of Technology (VIT), Chennai, Tamil Nadu, IndiaCentre for Smart Grid Technologies, School of Electrical Engineering, Vellore Institute of Technology, Chennai, IndiaImage processing is an essential first step towards fully utilizing robotics, deep learning, and machine learning techniques. Using techniques like image enhancement, restoration, and segmentation are able to extract pertinent information from images and use it for task execution and decision-making. However, the hardware implementation of these algorithms demands more delay, area, and power. This work proposes a new mantissa bit-size reduced half-precision floating-point format for processing and characterizing image pixels for machine learning algorithms. In the realm of imaging, lowering the mantissa bit-size in floating-point can conserve area and power when utilized for internal calculations. Together with the area-power reduction, there is also a progressive reduction in image quality. For any image application, it becomes necessary to monitor the area and power trade-offs related to the amount of bits used to process the raw data. This work assists in selecting the bit-size for internal computations based on the accuracy requirements of the application by reporting the area, and power for various bit size reductions. The updated pixel values after applying the mantissa bit-size reduction are displayed in this study, along with a theoretical explanation of its inaccuracy. Since multipliers and adders are needed for the majority of mathematical equations in machine learning image algorithms, they are developed later in this work to process the image. The processed image is based on various adjusted pixel values, and the experimental findings demonstrate 75.2% to 21.3% area optimization, and 66.43% to 20.4% power optimization. However, determining the PSNR and MSE values of the processed image allows for quality validation.http://www.sciencedirect.com/science/article/pii/S2590123024014348Half-precision floating-pointPixel characterizationBit-size reductionError analysisImagingAI applications |
| spellingShingle | J. Jean Jenifer Nesam S. Sankar Ganesh Sitharthan Ramachandran Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications Results in Engineering Half-precision floating-point Pixel characterization Bit-size reduction Error analysis Imaging AI applications |
| title | Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications |
| title_full | Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications |
| title_fullStr | Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications |
| title_full_unstemmed | Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications |
| title_short | Effect of bit-size reduced half-precision floating-point format on image pixel characterization for AI applications |
| title_sort | effect of bit size reduced half precision floating point format on image pixel characterization for ai applications |
| topic | Half-precision floating-point Pixel characterization Bit-size reduction Error analysis Imaging AI applications |
| url | http://www.sciencedirect.com/science/article/pii/S2590123024014348 |
| work_keys_str_mv | AT jjeanjenifernesam effectofbitsizereducedhalfprecisionfloatingpointformatonimagepixelcharacterizationforaiapplications AT ssankarganesh effectofbitsizereducedhalfprecisionfloatingpointformatonimagepixelcharacterizationforaiapplications AT sitharthanramachandran effectofbitsizereducedhalfprecisionfloatingpointformatonimagepixelcharacterizationforaiapplications |