US-20260127432-A1 - PROCESSING SENSING DATA USING DEDICATED ARTIFICIAL INTELLIGENCE PROCESSOR TO SWITCH BETWEEN POWER MODES

US20260127432A1US 20260127432 A1US20260127432 A1US 20260127432A1US-20260127432-A1

Abstract

A heterogeneous processor includes a first processor and a second processor of a different type. The heterogeneous processor operates in either a low-power mode or a full-power mode. The first processor is configured to operate in the low-power mode, process sensing data from a sensor using a trained neural network model, and generate a wake-up signal when an output of the trained neural network model satisfies a predefined criterion. The wake-up signal is provided to the second processor during the low-power mode. The second processor remains in a powered-down state during the low-power mode and transitions to the full-power mode in response to the wake-up signal.

Inventors

Lok Won Kim

Assignees

DEEPX CO., LTD.

Dates

Publication Date: 20260507
Application Date: 20251219
Priority Date: 20190104

Claims (20)

1 . A method of controlling an electronic device, comprising: placing the electronic device in a first state that consumes less power than a second state, a first processor of the electronic device in the first state consuming less power than in the second state; capturing sensing data in the first state by one or more sensors of the electronic device; generating a trigger signal by at least executing a neural network model on the sensing data by a second processor in the first state, the second processor configured to consume less power than the first processor in the second state; and switching the electronic device from the first state to the second state, responsive to generating the trigger signal.
2 . The method of claim 1 , wherein the first processor is a central processing unit (CPU) or an application processor (AP) and the second processor is an artificial intelligence (AI) acceleration processor.
3 . The method of claim 1 , wherein the first state comprises placing the electronic device in a stop mode, a sleep mode or a lock mode, and the second state comprises placing the electronic device in a boosting mode, an activation mode or an unlock mode.
4 . The method of claim 1 , further comprising sending the trigger signal from the second processor to the first processor, the electronic device switching from the first state to the second state responsive to receiving of the trigger signal by the first processor.
5 . The method of claim 1 , wherein the one or more sensors comprise a camera that generates image data as the sensing data.
6 . The method of claim 5 , wherein the neural network model is trained to perform facial recognition on the image data.
7 . The method of claim 1 , wherein the first processor and the second processor are integrated on a same system-on-chip (SOC).
8 . The method of claim 7 , wherein the sensing data is received directly from the one or more sensors by the SOC in real-time without storing the sensing data in memory outside the SOC.
9 . The method of claim 1 , wherein generating the trigger signal comprises: performing inference on the sensing data using the neural network model to generate determination data as an output of the neural network model; and comparing the determination data to reference data to determine whether the determination data meets or exceeds a threshold for switching to the second state.
10 . The method of claim 1 , wherein the neural network model is configured to process the sensing data of multiple modality.
11 . An electronic device, comprising: a camera configured to capture one or more images, the camera turned on in a first mode consuming less power than in a second mode; a dedicated artificial intelligence (AI) acceleration processor, during at least the first mode, configured to: receive the one or more images from the camera without storing the one or more images in a persistent file system, and generate wake-up data by at least processing the one or more images using a trained neural network model to detect a predetermined condition, a general processor; and a power source unit configured to: in the first mode: supply power to the camera and the dedicated AI acceleration processor but not the general processor, in the second mode: supply power to the camera, control unit, and the general processor, and switch from the first mode to the second mode in response to receiving the wake-up data from the dedicated AI acceleration processor.
12 . The electronic device of claim 11 , further comprising a microphone configured to capture one or more voices, wherein the power source unit is further configured to supply power to the microphone in the first and second mode, and wherein the wake-up data is generated from the one or more images or the one or more voices.
13 . The electronic device of claim 11 , wherein the trained neural network model includes a voice recognition neural network model and an image recognition neural network model.
14 . The electronic device of claim 11 , wherein the predetermined condition comprises detecting, in the first mode, an optical pattern from which the wake-up data is extracted.
15 . The electronic device of claim 11 , wherein the trained neural network model is configured to process heterogeneous sensing data.
16 . The electronic device of claim 11 , wherein the wake-up data comprises a binary signal transmitted to the control unit.
17 . A specialized processor in an electronic device, the specialized processor comprising: a dedicated artificial intelligence (AI) acceleration processor configured to, in a lock mode of the electronic device: receive an image from a camera or a voice from a microphone, generate wake-up data for transitioning the electronic device to an unlock mode, the wake-up data generated by processing the image or the voice using a trained neural network model to detect a predetermined condition, wherein power is selectively supplied to the microphone, the camera, and the dedicated AI acceleration processor in the lock mode, wherein power is supplied to the microphone, the camera, the dedicated AI acceleration processor and a general processor in the unlock mode based on the wake-up data, and wherein power consumption of the electronic device is less in the lock mode than in the unlock mode.
18 . The specialized processor of claim 17 , wherein the electronic device is a smart device including a smart phone, a computer, a home appliance, or a vehicle.
19 . The specialized processor of claim 17 , wherein the trained neural network model is configured to process heterogeneous sensing data including the image and the voice.
20 . The specialized processor of claim 17 , wherein the predetermined condition comprises detecting an optical pattern from which the wake-up data is extracted.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application is a continuation of U.S. Patent Application No. 19/276,003 filed on July 22, 2025, which claims priority to U.S. Patent Application No. 18/918,137 filed on October 17, 2024, which is a continuation of U.S. Patent Application No. 17/870,529 filed on July 21, 2022 (issued as U.S. Patent No. 12,165,061 on December 10, 2024), which is a continuation of U.S. Patent Application No. 17/366,042 filed on July 2, 2021 (issued as U.S. Patent No. 11,429,180 on August 30, 2022), which is a bypass continuation of International PCT Application No. PCT/KR2019/012420 filed on September 24, 2019, which claims the benefit of priority to Korean Application No. 10-2019-0001406 filed on January 4, 2019 and Korean Application No. 10-2019-0002220 filed on January 8, 2019, which are incorporated by reference herein their entirety. BACKGROUND OF THE DISCLOSURE TECHNICAL FIELD The present disclosure relates to a trained model creation method for performing a specific function for an electronic device, a trained model for performing a specific function for an electronic device, a dedicated chip for performing a specific function for an electronic device, an operation method for a dedicated chip for performing a specific function for an electronic device, an electronic device having a function of performing a specific function, and a system for performing a specific function of an electronic device, and more particularly, to a trained model creation method for performing a specific function for an electronic device, a trained model for performing a specific function for an electronic device, a dedicated chip for performing a specific function for an electronic device, an operation method for a dedicated chip for performing a specific function for an electronic device, an electronic device having a function of performing a specific function, and a system for performing a specific function of an electronic device for performing a specific function, which is fast and accurate, using a model which is trained in advance using an artificial neural network for an electronic device. BACKGROUND ART In the case of an electronic device such as a smart phone, power for all the unused hardware components is turned on even when the user does not use them, thereby a lot of power consumption is caused. To solve this problem, there has been an effort to reduce unnecessary power consumption by turning off the power for the unused hardware components when the user does not use them. In spite of this effort, according to a specific function performing system of a contemporary art, there is a technical limit in that a sensor cannot precisely recognize sensing data so that a specific function is performed in a situation in which the sensing data does not need to be sensed or a specific function is not performed even in a situation in which the sensing data needs to be sensed. SUMMARY OF THE DISCLOSURE The present disclosure is directed to solving the above-mentioned problem and an object of the present disclosure is to perform a specific function in an exact situation intended by a user by precisely understanding the sensing data. Further, an object is to more quickly and precisely output determination data of a specific function to be performed, by inputting sensing data to an AI recognition model. Further, an object is to promote the convenience of users only by performing an inference process by an AI recognition model without performing separate learning whenever real-time sensing data is inputted, using a previously trained AI recognition model to output determination data of a specific function to be performed. Finally, it is advantageous in that the power is not always turned on, but the system is driven only when specific sensing data is received to reduce power consumption. One aspect of the present disclosure provides a trained model creation method for performing a specific function for an electronic device, including: preparing big data for training an artificial neural network including, in pairs, sensing data received from a random sensing data generation unit for sensing human behaviors and determination data of performing specific function for determining whether to perform a specific function of an electronic device with respect to the sensing data; preparing an artificial neural network model, which includes nodes of an input layer through which the sensing data is inputted, nodes of an output layer through which the determination data of performing specific function of the electronic device is outputted, and associated parameters between the nodes of the input layer and the nodes of the output layer and calculates inputs of the sensing data for the nodes of the input layer in order to output the determination data of performing specific function from the nodes of the output layer; and repeatedly performing a process of inputting the sensing data included in the prepared big data in