Reducing processing load using transfer learning and learning models! A case of achieving automatic recipe suggestions.
The customer had the challenge of wanting to "implement image recognition technology that recognizes ingredients with a camera, and after correctly identifying the ingredients, propose recipes linked to the dishes." In implementing the image recognition technology, it was necessary to accurately recognize the materials with the camera, and to reduce processing load, transfer learning and a learning model (mobilenet) were used to ensure it could also operate on small PCs like Raspberry Pi. In recognizing work processes, the system was designed to recognize the work status at each stage with a camera, identify how far along the process is, and determine what the next step is, including follow-up to the next stage. Additionally, there are examples of product control through gestures. For more details, please refer to the related links below. [Case Overview (Partial)] ■ Client Industry: Electronics ■ Service Line: Technology / IoT *For more details, please refer to the PDF materials or feel free to contact us.
Inquire About This Product
basic information
【Summary of Other Cases】 <Challenges Faced by the Customer> ■ Concept Proposal: A - Implement image recognition technology to recognize ingredients with a camera, and after correctly identifying the ingredients, propose recipes linked to the dishes. ■ Concept Proposal: B - Control various products through gesture recognition. <Support Content and Results> ■ Realization of "Concept Proposal: A" - Achieved processing load reduction using transfer learning and a learning model (mobilenet). - Enabled the camera to recognize the work status at each step in the work process, identifying how far along the process is and what the next step is, including follow-up to the next step. ■ Realization of "Concept Proposal: B" - Improved recognition accuracy through image recognition technology and machine learning (excluding the effects of ambient light). - Processed image recognition and weight sensor information on the device side, integrated the information with the cloud for app connectivity, and established a system that allows users to control it. *For more details, please refer to the PDF document or feel free to contact us.
Price range
Delivery Time
Applications/Examples of results
For more details, please refer to the PDF document or feel free to contact us.
catalog（1）
Download All Catalogs
Company information
Pactera Technology Japan is a member of the Pactera Group, which has numerous locations globally and provides digital consulting and technology services to clients. Leveraging its accumulation of cutting-edge technology, it strongly promotes system construction that is organically combined with a hybrid application development system for clients across various industries, from finance to manufacturing and telecommunications. Currently, the Asia-Pacific region is a strategically important area for the group, and it is focusing on investing in and strengthening digital development, IoT solutions, and Fintech in that region. Additionally, Pactera offers various products and solutions to improve the operations of those in the manufacturing industry, as well as outsourced development services provided by professionals well-versed in manufacturing.
![[Book] Development of Biological Sensing Technology (No. 2245)](https://image.www.ipros.com/public/product/image/d02/2001189446/IPROS05038075950002086345.gif?w=280&h=280)
![Development of technologies related to early healing of fractures [Public Suwa University of Science and Technology]](https://image.www.ipros.com/public/product/image/42e/2001438525/IPROS30830840754359713339.png?w=280&h=280)
![Simple hand washing station "Oku-sen" [No plumbing work required, ready to use as soon as it arrives!]](https://image.www.ipros.com/public/product/image/100/2000357537/IPROS07366987947003880157.jpeg?w=280&h=280)
![[Rental] SRS220A Headset / Earphone Microphone Set](https://image.www.ipros.com/public/product/image/254366/IPROS4827147386115905832.jpg?w=280&h=280)