CN111541777A

CN111541777A - Voice-assisted blind person object recognition system based on yolo algorithm

Info

Publication number: CN111541777A
Application number: CN202010447143.2A
Authority: CN
Inventors: 阮继盛
Original assignee: Individual
Current assignee: Individual
Priority date: 2020-05-25
Filing date: 2020-05-25
Publication date: 2020-08-14

Abstract

The invention discloses a voice-assisted blind person identification system based on a yolo algorithm, which comprises a photographing mechanism, a control mechanism and a host, wherein the photographing mechanism comprises a transverse plate, the top of the transverse plate is fixedly connected with a camera, the bottom of the transverse plate is fixedly connected with an elastic cloth ring, the control mechanism comprises an elastic finger sleeve, a light-touch switch is embedded in the elastic finger sleeve, and the host comprises a shell; according to the invention, by arranging the photographing mechanism, the control mechanism, the host and the like, when the blind-person-holding type elastic finger sleeve is used, the host can be fixed at the waistband of the blind person, in a backpack or in a trouser pocket by using the clamping plate, the photographing mechanism is sleeved at the wrist of the hand of the blind person for holding the blind stick by using the elastic cloth ring, the elastic finger sleeve is worn at the thumb or forefinger of the hand, when the blind person needs to identify a front object, the camera can be used for photographing by pressing the light-touch switch, the operation is simple, and the wearing is convenient.

Description

Voice-assisted blind person object recognition system based on yolo algorithm

Technical Field

The invention relates to an object recognition system, in particular to a voice-assisted blind person object recognition system based on a yolo algorithm.

Background

With the development of artificial intelligence technology, in recent years, technology for performing high-speed and high-precision identification on objects in images by adopting artificial intelligence has appeared, and most of the technology adopts an advanced algorithm of a convolutional neural network. The convolutional neural network is a neural network that is similar to a neural network including convolutional calculation and has a deep structure, and is one of representative algorithms for deep learning. The convolution characteristic is very suitable for the application in the field of computer images, and the YOLO (you only look once) algorithm is one of a plurality of algorithms for identifying and classifying objects by applying artificial intelligence, and has higher efficiency and accuracy. It employs a single convolutional neural network to predict multiple regions of interest and class probabilities. Compared with the traditional object monitoring method, the algorithm has higher speed, so that the method is very suitable for a voice-assisted blind person object-learning system.

The work flow of the voice-assisted blind person identification system is mainly to photograph an object through a camera, analyze and identify the object in a comparison sheet by using a microcomputer, and finally broadcast the voice by using a loudspeaker so that the blind person can fully know the surrounding objects.

However, the existing voice-assisted blind person identification system is high in cost and inconvenient to carry, the blind person generally needs to hold a blind stick with one hand and touch the other hand, the blind person is inconvenient to operate by holding equipment with the other hand, and the system is inconvenient to use.

Disclosure of Invention

The invention aims to provide a yolo algorithm-based voice-assisted blind person identification system to solve the problems in the background art.

In order to achieve the purpose, the invention provides the following technical scheme:

the utility model provides a supplementary blind person's system of becoming aware of based on voice of yolo algorithm, includes mechanism, control mechanism and the host computer of shooing, it includes the diaphragm to shoot the mechanism, the top fixedly connected with camera of diaphragm, the bottom fixedly connected with elasticity cloth ring of diaphragm, control mechanism includes the elasticity dactylotheca, the inside embedding of elasticity dactylotheca sets up and dabs the switch, the host computer includes the casing, the inside fixedly connected with battery of casing, the inside fixedly connected with microcomputer of casing, the through-hole has been seted up to the outer wall of casing, the inner wall of casing and the department's fixedly connected with speaker that corresponds of through-hole, camera, dabs the switch all with microcomputer electric connection.

As a further scheme of the invention: the both ends symmetry of diaphragm is provided with two curb plates, two the mounting groove, two have all been seted up to the bottom of curb plate the inner wall difference first pin pole of fixedly connected with and the second pin pole of mounting groove, the fixed band has been cup jointed in the outer wall rotation of first pin pole, one side outer wall of fixed band fixedly connected with child magic subsides and female magic subsides in proper order, the outer wall of second pin pole is walked around to the one end of fixed band.

As a further scheme of the invention: and two ends of the transverse plate are respectively hinged with the top ends of the two side plates.

As a further scheme of the invention: the outer wall fixedly connected with of casing articulates the seat, the tip of articulated seat articulates there is splint, articulated seat is provided with the torsional spring with the articulated department of splint.

As a still further scheme of the invention: the top of the shell is provided with an earphone interface in a penetrating mode, and the earphone interface is electrically connected with the microcomputer.

Compared with the prior art, the invention has the beneficial effects that:

1. when the blind person holding the blind stick, the camera points to the front of the blind person, the elastic finger sleeve is worn at the thumb or forefinger of the hand to facilitate the operation of the tact switch, when the blind person needs to identify an object in front, the camera can be used for taking a picture by pressing the tact switch, and then the identification result is broadcasted by using the loudspeaker after the image is analyzed and identified by the microcomputer.

2. According to the invention, the transverse plate, the side plates, the fixing belt, the elastic cloth ring and the like are arranged, the second pin rod can be used as a fulcrum to drive the two side plates to rotate with the transverse plate by pulling the end part of the fixing belt, so that the angle between the two side plates and the transverse plate is adjusted to adapt to the use of the blind with different wrist sizes, the photographing mechanism can be fixed at the wrist of the blind through the matching of the sub magic tape and the female magic tape, the elastic cloth ring can play a role in pre-fixing and protecting the wrist skin of a user, and the side plates and the transverse plate are prevented from hurting the skin.

3. When the mobile phone is used, a user can open a wireless hotspot of the mobile phone for connecting the microcomputer with the Internet, and obtains the support of Baidu voice API synthesis service through an REST API based on an HTTP request, after a camera finishes shooting, the microcomputer identifies a shot picture through a yolo algorithm to obtain name entries of main objects in the picture, converts the obtained name entry texts into playable voice files through the Baidu voice API synthesis service, and plays the voice files through a loudspeaker, so that the system is not limited to built-in audio files with certain quantity and length, and can flexibly play required sounds according to requirements.

4. The voice-assisted blind person identification system is designed in a split mode, so that the weight of objects worn by the hands of a blind person user can be greatly reduced, the influence on the flexibility of the hands of the blind person is avoided, the maintenance cost and the maintenance difficulty of the identification system are reduced, when a photographing mechanism, a control mechanism or a host fails, only the failure mechanism needs to be maintained or replaced independently, the use cost of the blind person is reduced, and the identification system can be better popularized in the market.

Drawings

Fig. 1 is a schematic structural diagram of a voice-assisted blind person identification system based on a yolo algorithm.

Fig. 2 is a schematic structural diagram of a photographing mechanism in a voice-assisted blind person identification system based on the yolo algorithm.

Fig. 3 is a schematic structural diagram of a control mechanism in a yolo algorithm-based voice-assisted blind person identification system.

Fig. 4 is a schematic structural diagram of a host in a yolo algorithm-based voice-assisted blind person identification system.

Wherein, mechanism 1 shoots, control mechanism 2, host computer 3, diaphragm 4, camera 5, curb plate 6, mounting groove 7, first pin 8, second pin 9, fixed band 10, son magic subsides 11, female magic subsides 12, elasticity cloth ring 13, elasticity dactylotheca 14, dab switch 15, casing 16, articulated seat 17, splint 18, microcomputer 19, battery 20, speaker 21, through-hole 22, earphone interface 23.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Referring to fig. 1 to 4, in the embodiment of the invention, a voice-assisted blind person identification system based on a yolo algorithm includes a photographing mechanism 1, a control mechanism 2 and a host 3, the photographing mechanism 1 includes a transverse plate 4, a camera 5 is fixedly connected to the top of the transverse plate 4, an elastic cloth ring 13 is fixedly connected to the bottom of the transverse plate 4, the control mechanism 2 includes an elastic finger stall 14, a light touch switch 15 is embedded in the elastic finger stall 14, the host 3 includes a housing 16, a storage battery 20 is fixedly connected to the interior of the housing 16, a microcomputer 19 is fixedly connected to the interior of the housing 16, a through hole 22 is formed in the outer wall of the housing 16, a speaker 21 is fixedly connected to a corresponding position of the inner wall of the housing 16 and the through hole 22, and the camera 5 and the light touch switch 15 are both electrically connected to the microcomputer 19.

The specific model of the microcomputer 19 is Raspberry Pi 3B +, which comprises a 64-bit quad-core CPU with a master frequency of 1.4GHZ, a 1GB memory, four USB2.0 interfaces, a Bluetooth module, a wifi module and a plurality of GPIO expansion interfaces, and the microcomputer is small in size and light in weight.

With the help of the WIFI module of the microcomputer 19, when the mobile phone is used, a user can open a wireless hotspot of the mobile phone to connect the microcomputer 19 with the internet, and obtain the support of the Baidu voice API synthesis service through an REST API based on an HTTP request, after the camera 5 finishes shooting, the microcomputer 19 identifies the shot photo through a yolo algorithm to obtain the name vocabulary entry of a main object in the photo, converts the obtained name vocabulary entry text into a playable voice file through the Baidu voice API synthesis service, and plays the voice file through a loudspeaker.

When the blind person holding the tactile stick is used, the photographing mechanism 1 is sleeved on the wrist of the hand of the blind person holding the tactile stick by the aid of the elastic cloth ring 13, the camera 5 is located on the back of the wrist, the camera 5 points to the front of the blind person when the blind person holds the tactile stick, the elastic finger sleeve 14 is worn on the thumb or forefinger of the hand, so that the light touch switch 15 is operated, when the blind person needs to identify an object in the front, the camera 5 can be used for photographing by pressing the light touch switch 15, and then after the image is analyzed and identified by the microcomputer 19, an identification result is broadcasted by the aid of the loudspeaker 21.

The both ends symmetry of diaphragm 4 is provided with two curb plates 6, and mounting groove 7 has all been seted up to the bottom of two curb plates 6, and the inner wall of two mounting grooves 7 is the first pin 8 of fixedly connected with and second pin 9 respectively, and the outer wall of first pin 8 rotates and has cup jointed fixed band 10, and one side outer wall of fixed band 10 is fixedly connected with sub-magic subsides 11 and female magic subsides 12 in proper order, and the outer wall of second pin 9 is walked around to the one end of fixed band 10.

The two ends of the transverse plate 4 are respectively hinged with the top ends of the two side plates 6.

When the blind person passes the hand through the elastic cloth ring 13 and ensures that the camera 5 is positioned at the back of the wrist, the second pin rod 9 can be used as a fulcrum to drive the two side plates 6 to rotate with the transverse plate 4 by pulling the end part of the fixing band 10, so that the angle between the two side plates 6 and the transverse plate 4 can be adjusted to adapt to the use of the blind persons with different wrist sizes, and the photographing mechanism 1 can be fixed at the wrist of the blind person by matching the sub magic tape 11 with the mother magic tape 12.

The outer wall fixedly connected with of casing 16 articulates seat 17, and the tip of articulating seat 17 articulates there is splint 18, and articulated seat 17 is provided with the torsional spring with the articulated department of splint 18, and through the cooperation of articulating seat 17, splint 18, torsional spring, usable splint 18 are fixed host computer 3 in blind person's waistband department, or in the trousers bag to prevent that the host computer from dropping and losing.

The top of casing 16 runs through and is provided with earphone interface 23, earphone interface 23 and microcomputer 19 electric connection, and through earphone interface 23's setting, the joinable earphone carries out voice broadcast to make the blind person user can hear clearly in noisy environment and report the content.

The working principle of the invention is as follows:

When the blind person passes the hand through the elastic cloth ring 13, and after the camera 5 is ensured to be positioned at the back of the wrist, the end part of the fixing band 10 is pulled, the second pin rod 9 can be used as a fulcrum to drive the two side plates 6 and the transverse plate 4 to rotate, so that the angle between the two side plates 6 and the transverse plate 4 can be adjusted, the fixing band is suitable for the blind persons with different wrist sizes, the wrist of the blind person can be fixed by the photographing mechanism 1 through the matching of the sub magic tape 11 and the mother magic tape 12, the elastic cloth ring 13 plays a role in pre-fixing, can play a role in protecting the wrist skin of a user, and avoids the side plates 6 and the transverse plate 4 from hurting the skin.

Through the cooperation of the hinge seat 17, the clamping plate 18 and the torsion spring, the host 3 can be fixed at the waistband of the blind, in a backpack or in a trouser bag by using the clamping plate 18, thereby preventing the host from falling and losing.

Through the setting of earphone interface 23, can connect the earphone and carry out voice broadcast to make blind person user can hear clearly in noisy environment and report the content.

According to the invention, through the split design of the voice-assisted blind person identification system, the weight of the object worn by the hand of the blind person user can be greatly reduced, so that the influence on the flexibility of the hand of the blind person is avoided, and meanwhile, the maintenance cost and the maintenance difficulty of the identification system are reduced.

Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that various changes in the embodiments and/or modifications of the invention can be made, and equivalents and modifications of some features of the invention can be made without departing from the spirit and scope of the invention.

Claims

1. The utility model provides a supplementary blind person's literacy system of pronunciation based on yolo algorithm, includes mechanism of shooing (1), control mechanism (2) and host computer (3), its characterized in that: mechanism (1) of shooing includes diaphragm (4), top fixedly connected with camera (5) of diaphragm (4), the bottom fixed connection of diaphragm (4) has elasticity cloth ring (13), control mechanism (2) include elasticity dactylotheca (14), the inside embedding of elasticity dactylotheca (14) sets up and dabs switch (15), host computer (3) include casing (16), the inside fixedly connected with battery (20) of casing (16), the inside fixedly connected with microcomputer (19) of casing (16), through-hole (22) have been seted up to the outer wall of casing (16), the inner wall of casing (16) and the department of correspondence fixedly connected with speaker (21) of through-hole (22), camera (5), dabbing switch (15) all with microcomputer (19) electric connection.

2. The yolo algorithm-based voice-assisted blind person knowledge system according to claim 1, wherein: the both ends symmetry of diaphragm (4) is provided with two curb plates (6), two mounting groove (7), two have all been seted up to the bottom of curb plate (6) the inner wall of mounting groove (7) is fixedly connected with first pin (8) and second pin (9) respectively, fixed band (10) have been cup jointed in the outer wall rotation of first pin (8), one side outer wall of fixed band (10) fixedly connected with sub-magic subsides (11) and female magic subsides (12) in proper order, the outer wall of second pin (9) is walked around to the one end of fixed band (10).

3. The yolo algorithm-based voice-assisted blind person identification system as claimed in claim 2, wherein: and two ends of the transverse plate (4) are respectively hinged with the top ends of the two side plates (6).

4. The yolo algorithm-based voice-assisted blind person knowledge system according to claim 1, wherein: the outer wall fixedly connected with of casing (16) articulates seat (17), the tip of articulated seat (17) articulates there is splint (18), articulated seat (17) is provided with the torsional spring with the articulated department of splint (18).

5. The yolo algorithm-based voice-assisted blind person knowledge system according to claim 1, wherein: an earphone interface (23) penetrates through the top of the shell (16), and the earphone interface (23) is electrically connected with the microcomputer (19).