JP2023044192A - Food product slice device and food product slice method - Google Patents

Abstract

To provide a food product slice device and slice method which detect the position and the size of a food product and can smoothly discharge the food product even when the food product is sliced on an inclined surface.SOLUTION: A food product slice device 100a includes a feed conveyor 20 and a discharge conveyor 40 arranged along a conveyance path of a food product M, a rotated slice blade 60 arranged between the feed conveyor 20 and the discharge conveyor 40, a food product photographing part 30 for photographing the food product M on the feed conveyor 20, a food product detection part 35 which is configured of multilayer neural network, and inputs an image and detects the position and the size of the food product M, and a control part 37 for controlling a feed amount on the basis of the position and the size of the food product M, wherein the food product photographing part 30 photographs a post image after the plurality of food products M are mounted on the feed conveyor 20 so as to be adjacent to each other, and the food product detection part 35 detects the positions and the sizes of the respective food products M.SELECTED DRAWING: Figure 2

Description

The present invention relates to a food slicing device and a food slicing method for slicing various types of food.

A food slicing device has a cutting device such as a slicing blade on the transport side of a transport conveyor that transports various foods, and slices the food transported by the transport conveyor.

For example, Patent Literature 1 (Japanese Unexamined Patent Application Publication No. 2004-90115) describes a food slicing device that can discharge food in an aligned state without losing the sliced shape even if the slicing angle is adjusted.
The food slicing device described in Patent Document 1 is a food slicing device in which a lower blade and a slicing blade that rotates in sliding contact with the lower blade and repeatedly crosses a food conveying path are arranged between a food feeding conveyor and a discharge conveyor. , the slicing blade and the lower blade are housed in a common blade housing case, and the blade housing case is disposed so as to be tiltable with respect to the feed conveyor. By changing the angle of the blade housing case with respect to the feed conveyor, the slicing angle of the food fed by the feed conveyor can be changed. Even if the angle of the blade containing case is changed, the positional relationship between the feed conveyor and the discharge conveyor does not change, and the sliced food can be delivered to the discharge conveyor without losing its shape.

Patent Document 2 (Japanese Unexamined Patent Application Publication No. 2009-178809) describes a material slicing method and a slicing apparatus that increase the degree of quantification of fish fillets.
The material slicing method and slicing device described in Patent Document 2 intermittently feeds the material to the cut receiving portion side, and while the material feeding is stopped, the material feed conveying path is moved close to the cut receiving portion. A method of slicing a material with a transverse cutting blade, in which the cut surface is photographed by a camera disposed toward the cut surface of the material at the discharge side position of the cut material, and the photographed image is set in advance The image area of the portion occupied by the color is measured, and the material feeding amount for each cutting is determined based on the value obtained by dividing the target cross-sectional area or its compensation value by the image cross-sectional area.

In addition, Patent Document 3 (Japanese Laid-Open Patent Publication No. 9-300285) describes a food slicer that accurately cuts both hard and soft foods to a desired thickness.
The food slicer described in Patent Document 3 is provided with a feed conveyor that feeds food by the thickness of the cut, a blade that passes through the cutting position on the front side of the feed conveyor is provided, and the food on the feed conveyor is fed and the front end of the food is cut. In a food slicer in which a blade passes through the cutting position each time it protrudes from the cutting position and the front end of the food projecting from the cutting position is cut by the blade, the blade includes a fixed blade fixed at the cutting position and a cutting position. It consists of a passing movable blade, the front end of the food projecting from the cutting position is sandwiched between the fixed blade and the movable blade, and the movable blade passes through the cutting position while being in contact with the fixed blade.

Furthermore, Patent Document 4 (Japanese Patent Application Laid-Open No. 2012-249606) describes a food cutting apparatus capable of suppressing the distortion of the cut surface of food to be cut, finishing the cut surface neatly, and reducing variations in cutting weight.
The food cutting device described in Patent Document 4 includes a feeding mechanism that intermittently conveys food, an upstream clamping mechanism that is arranged above the feeding mechanism and holds down the food, and a downstream side of the feeding mechanism. a cutting unit having a cutting blade for slicing food and a reciprocating means for reciprocating the cutting blade so as to be able to appear and retract in the transportation path of the food; In the food cutting apparatus having angle adjusting means for adjusting the angle, the cutting unit is arranged close to the downstream side of the cutting blade, and appears and disappears from the side opposite to the cutting blade across the food conveying path. 1. A downstream side clamping mechanism comprising a clamping portion for holding food and a clamp driving portion for moving the clamping portion toward and away from the food conveying path.

Further, detecting the size of food before processing the food is described in Patent Document 5 (Japanese Patent Application Laid-Open No. 2014-068640), for example.
The method for centering vegetable heads of Patent Document 5 is a conveyer belt having a tray for mounting the vegetable head, and when conveying the vegetable head to be cored, a measuring booth, a centering booth and a division are placed on the conveyor belt. The booths are arranged sequentially along the direction of travel, and in the measurement booth, the position of the core of the vegetable head and the size of the vegetable head itself are individually measured using an infrared sensor, and the obtained individual Data is calculated and supplied to a centering booth, and after removing the core of the headed vegetable in the centering booth based on the supplied data, the centered headed vegetable is sent to a division booth and divided into at least two or more. is divided into

Japanese Patent Application Laid-Open No. 2004-90115 JP 2009-178809 A JP-A-9-300285 JP 2012-249606 A JP 2014-068640 A

In a food slicing device such as pork cutlet, the food on the feed conveyor is usually sliced into a plurality of slices and discharged to the discharge conveyor, but the last slice of the food is not discharged smoothly, for example, the slice turns over. It may be ejected in a state or tilted. In particular, when the food is sliced on a surface that is inclined with respect to the thickness direction, the last sliced piece of the food that is conveyed is often not discharged smoothly. In this case, it is necessary for the operator to collect all the slices for each food item and return the slices to their original alignment. However, since the last undischarged slice remains near the rotating slicing blade, there is a problem that the operator cannot remove it by hand due to danger.
Therefore, in order to smoothly discharge the last slice of food, a plurality of foods are placed on the feed conveyor in a state adjacent to each other in the conveying path direction, and are placed downstream of the plurality of foods in the conveying direction. It is desirable to extrude the last slice (offcut) of the sliced food product with the next food product to be sliced and discharge continuously.

On the other hand, the food slicing device needs to divide the food according to the size of the food. In addition, when actually slicing food with a slicing blade, it is necessary to detect the position of the food on the feed conveyor and determine an appropriate feeding amount of the food.
For this purpose, it is conceivable to take an image of the food on the feed conveyor and detect the position and size of the food from the taken image. In the case of color differences, for example, it is difficult to detect the boundaries between each food item, even though the boundaries between the food items and other parts can be easily detected.

Also, if the shape of the food is substantially rectangular, the area of each slice can be made substantially the same by slicing the food at regular intervals. However, foods such as tonkatsu may have a substantially triangular shape, or may have a narrow width at one end. In such a case, if the food is sliced at even intervals, the area of each sliced piece will vary greatly.
When a skilled worker slices food with a kitchen knife, the slicing position may be adjusted according to the shape of the food. rice field.

The food slicing device described in Patent Document 1 is provided with a pressing conveyor, and the food is sandwiched between the feed conveyor and the pressing conveyor, thereby smoothly discharging the final slice of the food.
However, when the food is sliced on a surface that is inclined with respect to the thickness direction, the end of the holding conveyor on the slicing blade side is separated from the end of the feed conveyor in the conveying path direction, so the food is sliced without adjacent food. In the case of slicing, there is a problem that the final slice of food cannot be discharged smoothly.
In addition, the food slicing device described in Patent Document 1 does not describe means for detecting the position and size of the food.

In the slicing apparatus described in Patent Document 2, the slice cross section of the food to be sliced is photographed, and the feeding amount is controlled according to the slice cross section area, thereby increasing the degree of quantification of fish fillets. However, Patent Document 2 does not describe means for smoothly discharging the last slice of food.
In addition, the slicing apparatus described in Patent Document 2 is suitable in terms of increasing the degree of quantification of fish fillets, but it is necessary to photograph the slice cross section each time slicing to determine the feed amount. There is also the issue that it takes time to

The slicing devices described in Patent Documents 3 and 4 are inventions of slicing devices that accurately cut both hard and soft foods to a desired thickness or finish the cut surface neatly. Document 4 does not describe means for smoothly ejecting the last slice of food or detecting the position and size of the food to be cut.

In the method for removing the core of the headed vegetable described in Patent Document 5, it is described that the position of the core of the headed vegetable and the size of the headed vegetable itself are individually measured using an infrared sensor in a measurement booth. there is
In the measurement method described in Patent Document 5, it is possible to detect the position and size of food if the difference in brightness between the object to be measured (food) and the background is clear. It is difficult to detect food-food boundaries when they are placed adjacent to each other in a direction.

The main object of the present invention is to detect the position and size of the food and automatically control the feed amount of the food to be sliced. To provide a food slicing device and a food slicing method capable of smoothly discharging the food including the last sliced piece without changing the slicing posture.
Another object of the present invention is to provide a food slicing apparatus and a food slicing method that can make the area of each slice piece substantially the same even if the shape of the food is not substantially rectangular.
Another object of the present invention is to provide a food slicing apparatus and a food slicing method capable of slicing each food separately even if a plurality of types of food are mixed.

(1)
A food slicing device according to one aspect is a food slicing device that slices food into a plurality of slices in a direction that intersects the direction in which food is conveyed while conveying a plurality of foods, and is arranged along a food conveying route. a feed conveyor and a discharge conveyor, a lower blade and a rotating slicing blade disposed in a gap formed between the feed conveyor and the discharge conveyor, and a plurality of food items on the feed conveyor; a food imaging unit for photographing, a food detection unit configured by a multi-layered neural network and detecting one or more foods from an image, a control unit for controlling the feeding amount of the food based on the position and size of the food, , the multi-layer neural network is configured to detect the region and class of the object, and the food imaging unit captures images of a plurality of foods placed adjacent to each other in the conveying path direction on the feed conveyor. A multi-layer neural network identifies the position and size of each food item from a plurality of adjacently placed food items.

In the food slicing device according to one aspect, the position and size of the food are detected from the image of the food photographed by the food photographing unit, and based on the detected position and size of the food, the slicing blade and the position where the food should be sliced are determined. The distance between the two is calculated, the feed conveyor is controlled to sequentially feed the food by the calculated distance, and the food can be sliced by the slicing blade.
In addition, the food is sliced while a plurality of foods are placed adjacent to each other in the conveying path direction on a feed conveyor, and the last sliced piece is pushed out by the food to be sliced next, and is discharged. It is necessary to discharge smoothly without breaking the sliced form including the sliced piece. In particular, when the slicing blade is angled, the food cannot be sufficiently pressed down by the pressing conveyor, so the next food to be sliced pushes it out. However, in this case, since a plurality of food items are adjacent to each other in the conveying route direction, it becomes difficult to detect the position and size of each food item.
In the food slicing device according to one aspect, multiple food products are sliced together by using a multilayer neural network in the food detection unit that can directly predict the coordinates and sizes of each of multiple objects in one image. It allows detection of the position and size of each food item, even when placed next to each other. A multi-layered neural network of the food detector identifies the location and size of each food item from a group of adjacent food items. The food detection unit can simultaneously perform a plurality of identifications, and may simultaneously detect the position and size of each food from a plurality of adjacent food groups.
In addition, for example, when there are three adjacent foods that are photographed at one time, an image is photographed each time the food moves on the feed conveyor by one piece, and only the position and size of the food in the center of each image is photographed. may be detected. Alternatively, the positions and sizes of a plurality of food items may be detected from one image, and an image may be captured each time the food item moves by the detected number.
Alternatively, the food on the feed conveyor may be photographed as a moving image, and detected continuously from the photographed moving image.
In addition, since a plurality of food items are placed adjacent to each other in the transport path direction, the food items can be successively placed, thereby improving work efficiency.
A multi-layer neural network used in the food detection unit may be any one that can detect the region and class of each object from an image containing a plurality of objects. For example, Mask R-CNN or the YOLO system (you only look once) can be used.

(2)
A food slicing apparatus according to a second aspect of the present invention is a food slicing apparatus according to one aspect, further comprising a holding conveyor above the feeding conveyor, and for slicing the food, the food can be fed while sandwiched between the feeding conveyor and the holding conveyor. good.

In this case, by stably pressing the food regardless of the thickness of the food, the precision of the slicing position of the food can be improved.

(3)
A food slicing device according to a third aspect of the invention is the food slicing device according to one aspect or the second aspect of the invention, wherein the slicing blade and the lower blade are housed in a blade housing case, and the blade housing case extends in the conveying direction of the feed conveyor. On the other hand, the inclination angle with respect to the vertical direction may be arranged so as to be adjustable.

In this case, by adjusting the inclination angle of the blade housing case, the angle of the sliced surface of the food with respect to the direction of the conveying path can be easily adjusted.
If the tilt rotation center of the blade housing case is positioned near the cutting edge of the lower blade, the positional relationship between the cutting edge of the lower blade and the feed conveyor will remain unchanged even if the tilt angle of the blade housing case is changed. is almost unchanged, the food can be smoothly delivered to the discharge conveyor without losing its sliced form, which is preferable.

(4)
A food slicing device according to a fourth aspect of the present invention is a food slicing device according to the first aspect to the third aspect of the invention, wherein the food may be a processed food in which batter is applied around a sheet-like piece of meat.

In this case, the processed food can be automatically and quickly sliced according to the specifications, and can be smoothly delivered to the discharge conveyor without breaking the sliced shape. Examples of processed foods include pork cutlet, chicken cutlet, and croquette.

(5)
A food slicing device according to a fifth aspect of the invention is the food slicing device according to the first aspect to the third aspect of the invention, wherein the food may be a substantially sheet-like meat product.

In this case, the substantially sheet-shaped meat product can be automatically and quickly sliced according to the specifications, and can be smoothly delivered to the discharge conveyor without breaking the sliced shape.

(6)
A food slicing device according to a sixth aspect of the invention is the food slicing device according to the first aspect to the fifth aspect of the invention, wherein the teacher data for the multi-layer neural network is prepared as separate teacher data for each type of food, and the food detection unit may specify the type of food in addition to the location and size of the food, and the control unit may control the feeding amount of the food based on the location, size and type of the food.

As a result, even when a plurality of types of food are mixed and put into the food slicing device, the food can be sliced by a slicing method preset for each food.
For example, when slicing a large number of tonkatsu and croquettes to be put into a lunch box for sale, a conventional food slicing device first sets the food slicing device for tonkatsu, slices the required number of tonkatsu, and then changes the settings of the food slicing device to croquettes. and had to slice the croquettes.
In this case, the lunch box cannot be completed until the sliced croquette is ready, so a space for temporarily placing the sliced pork cutlet is required.
On the other hand, in the food slicing apparatus according to the sixth invention, for example, by alternately throwing pork cutlets and croquettes into the slicing apparatus, the sliced pork cutlets and croquettes discharged from the slicing apparatus are put into lunch boxes. , the lunch box can be completed sequentially, and the space for temporarily placing the sliced pork cutlet is not required.

(7)
A food slicing apparatus according to a seventh aspect is the food slicing apparatus according to the first aspect to the fifth aspect, wherein the teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food, and the food detection unit specifies the type of food in addition to the position and size of the food, and the control unit may notify when the type of food is different from the set food.

In this case, for example, if chicken cutlet is set as the type of food and the food that is actually photographed is fried chicken, the operator can be notified that the setting of the food is incorrect. As a method of notifying the operator, for example, screen display or notification by buzzer sound can be used.

(8)
A food slicing apparatus according to an eighth aspect of the invention is the food slicing apparatus according to the first aspect to the fifth aspect of the invention, wherein a plurality of types of teacher data for the multi-layer neural network are prepared as separate teacher data for each shape of the food, and the food is detected. The unit may detect the shape of the food in addition to the position and size of the food, and the control unit may control the feeding amount of the food based on the position, size and shape of the food.

When the food detection unit detects the position and size of the food and slices the food based on the detection results, the length of the sliced food in the conveying path direction becomes uniform. However, foods such as tonkatsu may have a substantially triangular shape, or may have a narrow width at one end. In such a case, if the food is sliced at even intervals, the area of each sliced piece will vary greatly.
When a skilled worker slices food with a kitchen knife, the slicing position may be adjusted according to the shape of the food. rice field.
In the food slicing device according to the eighth invention, a plurality of types of separate training data are prepared for each food shape, and the training data are used to learn a multi-layer neural network, thereby allowing the food detection unit to detect the shape of the food. is similar to the shape of training data. Then, for example, when it is detected that the photographed food is similar to the teacher data in which the width at one end is narrow, the slice interval at one end is shifted to the other end. It is possible to increase the satisfaction of the customer who purchases the food by controlling it to be larger than the food.

(9)
A food slicing device according to a ninth aspect of the present invention is the food slicing device according to one aspect to the eighth aspect of the invention, further comprising a cut surface imaging unit and a cut surface detection unit arranged on the food discharge side, The imaging unit photographs the cut surface of the food, and the cut surface detection unit calculates the area of the cut surface and/or the ratio of the fat portion and the lean portion using the photographed image of the cut surface, and determines the feed amount of the food. may be adjusted according to the area of the cut surface and/or the ratio of the fat portion to the lean portion.

For example, when cutting pork cutlet with a food slicing device, the cut surface of the pork cutlet is photographed, and if there is a lot of fat and little red meat, the amount of food fed is increased to equalize the weight of the lean meat for each slice. be able to. Also, the feeding amount of the food may be adjusted according to the area of the cut surface so that the weight or size of each slice may be uniform, or nutritional factors such as calories may be uniform. Also, by photographing the cut surface of the food M, the pressing force or height of the pressing conveyor may be changed.
Also, it is possible to determine whether the food is spoiled by the color, shape, etc. of the photographed image of the cut surface of the food. Therefore, the food slicing device may be added with a function of notifying the operator by a buzzer sound or the like when the food is spoiled.

(10)
A food slicing method according to another aspect includes a placing step of placing a plurality of foods on a feed conveyor adjacent to each other in the direction of a conveying route, a photographing step of photographing images of the plurality of foods, and multi-layering of image data. A detection process of inputting to a neural network and detecting the position and size of the food from the image, a feed control process of controlling the feeding amount of the food based on the position and size of the food, and a slicing blade that rotates and crosses the conveying path a slicing step of slicing the food product with a multi-layer neural network configured to detect regions and classes of objects, the detecting step comprising: Or detect the positions and sizes of a plurality of food items.

In this case, the food is sliced while the food is placed adjacent to each other in the conveying path direction on the feed conveyor, and the last sliced piece is pushed out by the food to be sliced next and discharged, whereby the final food is discharged. It is possible to smoothly discharge a plurality of sliced pieces including the sliced piece without breaking the sliced shape.
In addition, by using a multi-layer neural network capable of detecting regions and classes of objects, it is possible to detect the position and size of food items from an image in which a plurality of food items are placed adjacent to each other.

(11)
A food slicing method according to an eleventh aspect of the present invention is a food slicing method according to another aspect, wherein the teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food, and the detection step includes detecting the position of the food and In addition to size, the type of food item may be specified and the feed control step may control the amount of food item feed based on the location, size and type of food item.

As a result, even when a plurality of types of food are mixed and put into the food slicing device, the food can be sliced by a slicing method preset for each food.
For example, when slicing a large number of tonkatsu and croquettes to be put into a lunch box for sale, a conventional food slicing device first sets the food slicing device for tonkatsu, slices the required number of tonkatsu, and then changes the settings of the food slicing device to croquettes. and had to slice the croquettes.
In this case, the lunch box cannot be completed until the sliced croquette is ready, so a space for temporarily placing the sliced pork cutlet is required.
On the other hand, in the food slicing method according to the eleventh aspect of the present invention, for example, the pork cutlet and the croquette are alternately put into the slicing device, and the sliced pork cutlet and the croquette discharged from the slicing device are put into the lunch box. , the lunch box can be completed sequentially, and the space for temporarily placing the sliced pork cutlet is not required.

(12)
A food slicing method according to a twelfth aspect of the present invention is a food slicing method according to another aspect, wherein the teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food, and the detection step includes detecting the position of the food and In addition to the size, the type of food may be specified and the feed control process may notify if the type of food differs from the set food.

In this case, for example, if chicken cutlet is set as the type of food and the food that is actually photographed is fried chicken, the operator can be notified that the setting of the food is incorrect. As a method of notifying the operator, for example, screen display or notification by buzzer sound can be used.

1 is a schematic perspective view of the food slicing device of the first embodiment; FIG. It is explanatory drawing of imaging|photography of foodstuffs, and conveyance to the slicing blade side. FIG. 4 is an explanatory diagram of a state in which the slicing blade and the lower blade are tilted with respect to the feed conveyor; It is a schematic diagram which shows the structure of the control system of a food slicing apparatus. It is an image which shows an example of the detection result of foodstuffs, and a slice position. FIG. 6(a) is a flow chart showing an example of processing by the food imaging unit and the food detection unit, and FIG. 6(b) is a flow chart showing an example of processing by the control unit. FIG. 11 is a table showing an example of slicing parameters for each type of food in the second embodiment; FIG. It is a flow chart which shows an example of processing of a food photographing part in a 2nd embodiment, and a food detection part. It is an image which shows an example of the detection result of food in 3rd Embodiment, and a slice position. It is a flow chart which shows an example of processing of a food photography part in a 3rd embodiment, and a food detection part. It is a schematic diagram which shows the structure of food cutting surface vicinity of the food slicing apparatus of 4th Embodiment. It is a schematic diagram which shows the structure of the control system of the food slicing apparatus of 4th Embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description, the same reference numerals are given to the same parts. Moreover, in the case of the same reference numerals, their names and functions are also the same. Therefore, detailed description thereof will not be repeated.

[First embodiment]
FIG. 1 is a schematic perspective view of the food slicing device 100a of the first embodiment, FIG. 2 is an explanatory view of photographing the food M and conveying it to the slicing blade 60 side, and FIG. It is an explanatory view of a state in which the slicing blade 60 and the lower blade 65 are inclined with respect to each other.

1 to 3, the food slicing device 100a includes a feed conveyor 20, a discharge conveyor 40, a pressure conveyor 70, a food imaging unit 30, a slicing blade 60, a lower blade 65, a slicing blade 60 and and a blade housing case 50 for housing the lower blade 65 .
The food photographing unit 30 includes a camera 31 and an illumination device 32 and photographs images of a plurality of foods M placed on the feed conveyor 20 .
As shown in FIG. 2, a gap is formed between the feed conveyor 20 and the discharge conveyor 40, and the slicing blade 60 and the lower blade 65 are arranged in the gap. The slicing blade 60 is configured to rotate in sliding contact with the lower blade 65 and slice the food M across the transport path.

As shown in FIG. 1, the blade housing case 50 is composed of a case body 52 whose inclination angle can be adjusted with respect to the vertical direction, and a lid body 53 attached to the case body 52 via a hinge portion 58 so as to be openable and closable. be.
A lower blade 65 is replaceably fixed to the lower edge of the rectangular opening 54 formed in the case main body 52 by bolting or the like corresponding to the height of the transfer path of the feed conveyor 20 .
The lower blade 65 is a rectangular plate having a longitudinally elongated cross-section with an overall length about the width of the feed conveyor 20, and the upper edge on the downstream side (discharge conveyor 40 side) serves as a cutting blade.
The cutting edge is positioned near the center of inclination and rotation of the blade housing case 50 and is aligned with the height of the conveying path of the feed conveyor 20 . In the embodiment, the cutting edge of the lower blade 65 and the tilt rotation center of the blade housing case 50 are substantially aligned.

The cutting edge of the slicing blade 60 exhibits a curve similar to an involute curve. It makes sliding contact so that the contact length gradually increases, and traverses the conveying transition path once per rotation.
It should be noted that two cutting edges of the slicing blade 60 may be provided at mutually axially symmetrical positions so that the cutting edges traverse the conveying transition path twice for each rotation of the slicing blade 60 .

A guide base 68 is arranged in a gap formed between the feed conveyor 20 and the lower blade 65, and the guide base 68 is fixed at a fixed position regardless of the tilt adjustment of the blade housing case 50. - 特許庁
The upper portion of the guide table 68 is enlarged so that the upper surface serves as a guide surface.

A pressing conveyor 70 for pressing the food M placed on the feeding conveyor 20 is arranged above the feeding conveyor 20 on the downstream side. Although not shown in FIGS. 1 to 3, the support system and drive system of the pressure conveyor 70 stably press the food M regardless of the thickness of the food M, and also feed the food M at the same speed as the feed conveyor 20. is configured to send out For the structure of the support system and drive system of the pressing conveyor 70, for example, the structure described in Patent Document 1, which is a patent application filed by the present applicant, can be adopted.

As shown in FIG. 2, the foods M are placed adjacent to each other on the feed conveyor 20 in the conveying path direction. The placed food M is imaged by the food imaging unit 30 . The food imaging unit 30 can be arranged above the feed conveyor 20 . In this image capturing, for example, a total of three food items M, ie, the food items M in the center in the conveying direction and the food items M in front and behind thereof are captured as one image. More images of the food M may be taken depending on the length of the food M or the like.
The food M whose image has been photographed is sent in the downstream direction of the feed conveyor 20, is sandwiched between the feed conveyor 20 and the pressing conveyor 70, advances from the guide table 68 to a position crossed by the slicing blade 60, and is sliced there. The sliced pieces are pushed by the upstream remainder (sliced pieces) of the same food M or the next upstream food M to be sliced, and smoothly discharged to the discharge conveyor 40 .

When slicing food M such as pork cutlet, it is sometimes preferred that the slice plane is inclined with respect to the thickness direction of the food M. In such a case, it is necessary to incline the entire blade housing case 50 including the slicing blade 60 and the lower blade 65 with respect to the conveying direction of the feed conveyor 20 .
FIG. 3 shows the relationship between the slicing blade 60, the lower blade 65, the feed conveyor 20, the holding conveyor 70, etc. when the blade housing case 50 is inclined with respect to the conveying direction of the feed conveyor 20. FIG. By inclining the slicing blade 60 and the lower blade 65, the slicing surface of the food M can be inclined.

In this case, since the upper portion of the blade housing case 50 is tilted rightward in FIG. That is, the end portion of the holding conveyor 70 in the conveying direction is arranged upstream (right side in FIG. 3) of the end portion of the feed conveyor 20 in the conveying direction. When the pressing conveyor 70 is retracted rightward, the position where the food M is sandwiched between the feeding conveyor 20 and the pressing conveyor 70 moves rightward. In this case, the last slice (the most upstream slice) of the sliced food M cannot be discharged smoothly because it is not held down by the pressing conveyor 70 .
Therefore, in order to collect all the slices of the food M and display them in an aligned state, the food M is placed adjacent to each other on the feed conveyor 20, and is pushed out by the next food M in a continuous manner. It is preferable to let it drain. As a result, the food M is pushed out to the food M to be sliced next, so that all the sliced pieces are smoothly discharged to the discharge conveyor 40 .

FIG. 4 shows the configuration of the control system of the food slicing device 100a. The image data photographed by the food photographing unit 30 is sent to the food detection unit 35, and the position and size of the food M is detected by the food detection unit 35 for the food M in the center of the image of the three adjacent food M images. is detected, and the feed amount to each slice position of the food M is calculated. The feed amount of the food M to each slice position is sent to the control unit 37, and the control unit 37 instructs the feed conveyor 20 and the holding conveyor 70 to move forward corresponding to the feed amount, and the food M is fed to a predetermined position. The slicing blade 60 slices the food M at this timing.
At the timing of slicing the food M, the food M can be sliced cleanly when it is stationary. 20 and the pressing conveyor 70 are synchronously and intermittently driven to move the food M by a required feed amount, and then the food M is sliced in a stationary state.
However, if priority is given to the efficiency of slicing, the transport may be driven continuously.

FIG. 5 shows an example of a detection result image captured by the food imaging unit 30 and detected by the food detection unit 35 . The food photographing unit 30 normally photographs a total of three food items M adjacent to each other in the conveying direction.
The food detection unit 35 inputs the image data to the multi-layer neural network, and the multi-layer neural network outputs the coordinates and size of the bounding box of each food M. The food detection unit 35 further divides the length of the bounding box of the food M in the center of the image by a predetermined number of divisions (6 in the case of FIG. 5) to detect the position to be sliced. Calculate the feed, which is the distance between the 60 positions. Therefore, in the case of the first embodiment, the width of the sliced pieces of the food M in the conveying direction is uniform.
Note that "P_Outlet" in FIG. 5 indicates what the food M is, and "0.91" indicates that the probability that the food M is "P_Outlet" is 0.91.

As shown in FIG. 5, when AI detects foods M from image data of a plurality of adjacent foods M, depending on AI, the position and size of each food M may not be detected accurately, or may be detected. However, it may take some time for the detection to first determine the area and then classify what it is.
However, in the present invention, a multi-layered neural network equipped with a regression problem approach that directly predicts the coordinates and size of an object is adopted in the food detection unit 35. The position and size of each food item M can be detected quickly and accurately.
A multi-layered neural network is employed in the food detection unit 35, and the position and size of the food M are detected by placing a plurality of food M on the feed conveyor adjacent to each other in the conveying path direction. To automatically control the feeding amount of the food M to be sliced, and to maintain the sliced posture including the last slice of the food M even when the food M is sliced on a surface inclined with respect to the thickness direction. It is now possible to discharge smoothly.

FIG. 6(a) shows a flowchart showing an example of the processing of the food imaging unit 30 and the food detection unit 35. As shown in FIG.
In FIG. 6(a), first, images of three adjacent food items M are normally photographed, and the photographed image data are sent to the food detection unit 35 (step S11).
Next, the sent image is input to a multi-layer neural network to detect the position and size (size of bounding box) of food M in the center (step S12).
Next, using the designated slicing interval of the food M or dividing the length of the food M in the conveying direction by the designated number of divisions, the position to be sliced is calculated, and the position to be sliced and the slicing blade 60 A feed amount, which is a distance from the position, is calculated (step S13). If the width of the last sliced food M is significantly smaller as a result of calculating the position to be sliced using the specified slice interval of the food M, the width of the last sliced food M is substantially uniform. The slice position may be adjusted so that
Next, the feeding amount of the food M is transmitted to the control section 37 (step S14).
Next, it is confirmed whether or not there is new food M, and if there is no new food M, the process ends (step S15).
The photographing and detection of the food M are performed each time the food M is moved on the feed conveyor 20 by one piece.
However, the positions and sizes of a plurality of foods M may be detected from one image, and an image may be captured each time the foods M move by the detected number.
Alternatively, the food M on the feed conveyor 20 may be photographed as a moving image, and the position and size of the food M may be continuously detected from the photographed moving image.

FIG. 6(b) shows a flowchart showing an example of the processing of the control unit 37. As shown in FIG.
In FIG. 6B, first, the feeding amount of the food M is received from the food detection unit 35 (step S16).
Next, it is confirmed whether or not there is a new feeding amount of the food M in the received feed amount data, and if not, the process ends (step S17).
If there is a feeding amount of new food M, the feed conveyor 20 and the pressing conveyor 70 are driven according to the feeding amount (step S18).
The food M is sliced by the slicing blade 60 (step S19).
If the sliced food M still has an unsliced feeding amount, the process returns to step S18, and if not, the process returns to step S16 (step S20).

The food M to be sliced is not limited, but processed food that can assume various shapes is suitable. Especially in the case of easily deformable processed foods, if they are arranged adjacent to each other on the feed conveyor 20, their boundaries may not be detected by the conventional detection method. Examples of processed foods include cutlets, croquettes, ham, bacon, tempura, and hamburgers. Of these, tonkatsu in which sheet meat is battered and fried, or substantially sheet-like meat products are preferable.

For example, Mask R-CNN, YOLO system (you only look once), etc. can be used as a multilayer neural network capable of detecting the region and class of each object from an image containing a plurality of objects.
Mask R-CNN simultaneously performs generic object detection and segmentation, and was announced by Kaiming He et al. (Facebook AI Research) at ICCV (International Conference on Computer Vision) 2017.
By using Mask R-CNN, it is possible to accurately determine the location of food item M at the pixel level rather than just enclosing a bounding box.
Also, the YOLO system is a highly accurate object detection algorithm capable of real-time processing, and was announced by Joseph Redmon et al. at CVPR (Computer Vision and Pattern Recognition) 2016. After that, YOLOv3 was announced in 2018.
By using the YOLO system, it is possible to ensure high detection accuracy in real time even when the line speed is high.

[Second embodiment]
The hardware configuration of the food slicing device 100b of the second embodiment is the same as the hardware configuration of the food slicing device 100a of the first embodiment. apply.
In the food slicing apparatus 100b of the second embodiment, a plurality of types of teacher data for the multi-layer neural network are prepared for each type of food M, and further, for each food M, the slicing parameter (slice interval or number of divisions, etc.) set.
FIG. 7 shows an example of slicing parameters for each type of food M in the second embodiment. The slicing parameter is, for example, the number of divisions when slicing the food M or the slicing interval. In the case of FIG. 7, the tonkatsu is sliced at intervals of 2 cm, the croquette is divided into two, and the hamburger is divided into three.

FIG. 8 shows an example of a flow chart of processing of the food imaging unit 30 and the food detection unit 35 in the second embodiment.
In the second embodiment, first, for example, images of three adjacent foods M are photographed, and the photographed image data are sent to the food detection unit 35 (step S11). Among them, it is determined which kind of food M it is (step S21).
Then, in the step of detecting the position and length of the central food item M (step S12) and calculating the feed amount to each slice position, the slice parameters of the corresponding food item M are referred to (for example, if the food item M is pork cutlet If so, the slice interval is set to 2 cm, and if the food M is a croquette, the slice interval is set to half the length of the food M, and the feeding amount to each slice position is calculated (step S32). ).
Other steps are the same as in the first embodiment. Also, the flowchart of the processing of the control unit 37 is the same as that of the first embodiment.
In this embodiment, an example is shown in which the feeding amount is adjusted according to the type of food M to change the slice interval or the number of divisions. The pressing force of the pressing conveyor 70 may be changed, and the slicing angle of the slicing blade 60 may be changed between M and hard food M.
As a result, a plurality of types of food M can be appropriately cut based on the slicing parameters, so that a plurality of types of food M can be mixed and fed into the food slicing device 100b.

(Modification of Second Embodiment)
In the second embodiment, a plurality of types of food M are mixed and put into the food slicing device 100b. For example, even when only one type of food M is put into the food slicing device 100b, if the type of food M set by the operator and the type of food M determined by the food detection unit 35 are different, the screen display or It is also possible to call the operator's attention with a buzzer sound or the like.

[Third embodiment]
The hardware configuration of the food slicing device 100c of the third embodiment is the same as the hardware configuration of the food slicing device 100a of the first embodiment. apply.
In the food slicing device 100c of the third embodiment, a plurality of types of training data for the multi-layer neural network are prepared for each shape of the food M. FIG. The shape of the food M is a substantially rectangular shape, a substantially triangular shape, or a shape with one end narrowed.

FIG. 9 shows an example of a shape in which the central food item M among the three adjacent food items M has a narrower width at one end. In FIG. 9, since the food items M on the right and left sides have a substantially rectangular shape, there is no problem even if they are sliced with a uniform width. The area and/or mass of the slice near the right edge is reduced. In such a case, a skilled operator often adjusts the slicing position according to the shape of the food M, but the conventional food slicing apparatus could only slice the food with a uniform width. Further, in the slicing apparatus described in Patent Document 2, the slicing width can be adjusted by photographing the slice cross section and controlling the feed amount according to the cross-sectional area of the cut surface. There was a problem that it takes time to slice because it is necessary to shoot.

In the food slicing device 100c of the third embodiment, several types of teacher data are prepared, such as substantially rectangular teacher data (P_cutlet1) and teacher data (P_cutlet2) that is tapered near the right end. Train a multilayer neural network. Then, the food detection unit 35 determines which teacher data shape the food M is closest to, and sets the slice position according to the shape.
For example, for a pork cutlet having a shape similar to P_cutlet2, the slicing position is adjusted according to the shape of the food M, such as by increasing the slicing interval near the right end, in the same way as a skilled worker. By making the area and/or mass substantially uniform, it is possible to increase the satisfaction of the customer who purchases the food item M. Also, by adjusting the slicing position, the slicing time does not become longer than that of the food slicing device 100a of the first embodiment.

FIG. 10 shows an example of a flow chart of processing of the food imaging unit 30 and the food detection unit 35 in the third embodiment.
In the third embodiment, usually three adjacent food items M are photographed and the photographed image data is sent to the food detection unit 35. After step S11, the shape of the central food item M is changed to P_cutlet1, P_cutlet2, etc. , which one of a plurality of types of training data prepared in advance is closest (step S31).
Then, in the step of calculating the feed amount to each slice position after step S12, which detects the position and length of the central food item M, for example, if the shape of the food item M is close to P_cutlet1, the slice intervals are made uniform. , and if the shape of the food M is close to P_cutlet2, after making adjustments such as increasing the slice interval near the right end, the feed amount to each slice position is calculated (step S32).
Other steps are the same as in the first embodiment.
Also, the flow chart of the processing of the control unit 37 is the same as that of the first embodiment.

FIG. 11 is a schematic diagram showing the configuration of the food slicing device 100d of the fourth embodiment in the vicinity of the cut surface of the food M. As shown in FIG. Also, FIG. 12 is a schematic diagram showing the configuration of the control system of the food slicing device 100d.
The food slicing device 100d of the fourth embodiment further includes a cut surface imaging unit 80 and a cut surface detection unit 85 in addition to the food slicing device 100a of the first embodiment. The cut surface photographing unit 80 is arranged at a position on the side of the food M being discharged, and photographs the cut surface of the cut food M. As shown in FIG. Then, the captured image of the cut plane is input to a cut plane detection unit 85 configured by conventional image processing (without using a multilayer neural network) or image processing using a multilayer neural network, and the area of the cut plane, and/or the proportion of fat and lean portions is calculated. Then, in the food slicing device 100d of the fourth embodiment, the feed amount of the feed conveyor 20 is adjusted according to the ratio of the fat portion and the lean portion. As a result, for example, when cutting pork cutlet, the cut surface of the pork cutlet is photographed, and if there is a lot of fat and little red meat, the feeding amount can be increased to make the weight of the lean meat even for each slice. . In addition, the feed amount of the feed conveyor 20 may be adjusted according to the area of the cut surface of the food M so that the weight or size of each slice may be uniform, or nutritional elements such as calories may be uniform. can be Also, the pressing force or height of the pressing conveyor 70 may be changed by photographing the cut surface of the food M.

Also, it is possible to judge whether the food M is spoiled or not from the color, shape, etc. of the photographed image of the cut surface of the food M. Therefore, the food slicing device 100d may be added with a function of notifying the operator by a buzzer sound or the like when the food M is spoiled.

In the present invention, food M corresponds to "food", food slicing devices 100a, 100b, 100c, and 100d correspond to "food slicing devices", feed conveyor 20 corresponds to "feed conveyor", and discharge conveyor 40 The lower blade 65 corresponds to the “lower blade”, the slicing blade 60 corresponds to the “slicing blade”, the food imaging unit 30 corresponds to the “food imaging unit”, and the food detection unit 35 corresponds to the "food detection unit", the control unit 37 corresponds to the "control unit", the holding conveyor 70 corresponds to the "holding conveyor", the blade housing case 50 corresponds to the "blade housing case", The cut plane imaging unit 80 corresponds to the "cut plane imaging unit", and the cut plane detection unit 85 corresponds to the "cut plane detection unit".

Although one preferred embodiment of the invention is described above, the invention is not so limited. It is understood that various other embodiments can be made without departing from the spirit and scope of the invention. Furthermore, in this embodiment, the actions and effects of the configuration of the present invention are described, but these actions and effects are examples and do not limit the present invention.

20 feed conveyor 30 food imaging unit 35 food detection unit 37 control unit 40 discharge conveyor 50 blade housing case 60 slicing blade 65 lower blade 70 holding conveyor 80 cut surface imaging unit 85 cut surface detection unit 100a, 100b, 100c, 100d food slice Apparatus M Food

Claims (12)

A food slicing device for conveying a plurality of foods and slicing the foods into a plurality of slice pieces in a direction intersecting the conveying direction of the food,
a feed conveyor and a discharge conveyor arranged along the transport path of the food;
a lower blade and a rotating slicing blade disposed in a gap formed between the feed conveyor and the discharge conveyor;
a food imaging unit that captures images of the plurality of foods on the feed conveyor;
a food detection unit configured with a multi-layer neural network for detecting one or more of the food items from the image;
a control unit that controls the feeding amount of the food based on the position and size of the food,
The multi-layer neural network is configured to detect regions and classes of objects,
The food photographing unit photographs the images of the plurality of foods placed adjacent to each other in the conveying path direction on the feed conveyor,
The food slicing device, wherein the multi-layer neural network specifies the position and size of each of the food items placed adjacent to each other. A holding conveyor is further provided on the upper part of the feed conveyor,
2. The food slicing apparatus according to claim 1, wherein the food is sandwiched between the feed conveyor and the holding conveyor for slicing. 2. The slicing blade and the lower blade are housed in a blade housing case, and the blade housing case is arranged such that an inclination angle with respect to a direction perpendicular to the conveying direction of the feed conveyor is adjustable. 3. The food slicing device according to 2. 4. The food slicing apparatus according to any one of claims 1 to 3, wherein the food is a processed food in which a sheet-like piece of meat is coated with a coating. The food slicing device according to any one of claims 1 to 3, wherein the food is a substantially sheet-like meat product. A plurality of teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food,
The food detection unit identifies the type of the food in addition to the position and size of the food,
The food slicing apparatus according to any one of claims 1 to 5, wherein the control unit controls the feeding amount of the food based on the position, size and type of the food. A plurality of teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food,
The food detection unit identifies the type of the food in addition to the position and size of the food,
The food slicing device according to any one of claims 1 to 5, wherein the control unit notifies when the type of the food is different from the set food. A plurality of types of teacher data for the multi-layer neural network are prepared as separate teacher data for each shape of the food,
The food detection unit detects the shape of the food in addition to the position and size of the food,
The food slicing apparatus according to any one of claims 1 to 5, wherein the control section controls the feeding amount of the food based on the position, size and shape of the food. Furthermore, a cut surface imaging unit and a cut surface detection unit arranged on the discharge side of the food are provided,
The cut surface photographing unit photographs the cut surface of the food,
The cut surface detection unit uses the photographed image of the cut surface to calculate the area of the cut surface and/or the ratio of the fat portion and the lean portion,
9. The food slicing device according to any one of claims 1 to 8, wherein the feeding amount of said food is adjusted according to said area and/or said ratio. A placing step of placing a plurality of foods on the feed conveyor adjacent to each other in the conveying path direction;
a photographing step of photographing a plurality of images of the food;
a detection step of inputting the data of the image into a multi-layer neural network and detecting the position and size of the food from the image;
a feed control step of controlling the feeding amount of the food based on the position and size of the food;
a slicing step of slicing the food with a slicing blade that rotates and traverses a conveying path;
The multi-layer neural network is configured to detect regions and classes of objects,
The food slicing method, wherein the detecting step detects the position and size of one or more of the food items from the image in which the multiple food items are placed adjacent to each other. A plurality of teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food,
The detecting step identifies the type of the food in addition to the position and size of the food,
11. The food slicing method according to claim 10, wherein said feeding control step controls the feeding amount of said food based on the position, size and type of said food. A plurality of teacher data for the multi-layer neural network are prepared as separate teacher data for each type of food,
The detecting step identifies the type of the food in addition to the position and size of the food,
11. The food slicing method according to claim 10, wherein said feeding control step notifies when the type of said food is different from said set food.

Images