CN109242183B - Crowd simulation evacuation method and device based on artificial fish-swarm algorithm and target detection - Google Patents

Abstract

The invention discloses a crowd simulation evacuation method and device based on artificial fish swarm algorithm and target detection. The method includes the following steps: establishing a topology map according to the plan of the evacuation scene, wherein the nodes represent evacuation exits, and the connecting lines represent two The exit is connected, and the final exit is marked; according to the video data, the individuals of the evacuated crowd are detected and tracked to obtain the real-time speed of the evacuated crowd; the individual speed is used to represent the feeding speed in the feeding behavior of the fish, initialize the fish, and use the scene The final exit position of , represents food, and the individual evacuated crowd is planned based on the artificial fish swarm algorithm; when the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is obtained. The invention identifies individuals with slow evacuation speeds, and guides crowds to evacuate, thereby taking care of vulnerable groups in evacuation, improving evacuation efficiency and avoiding excessive congestion.

Description

Crowd simulation evacuation method and device based on artificial fish swarm algorithm and target detection

technical field

The invention belongs to the field of crowd evacuation simulation, and in particular relates to a crowd simulation evacuation method and device based on artificial fish swarm algorithm and target detection.

Background technique

In the actual evacuation situation, the problems are becoming more and more complex, there are many non-deterministic and unknowable factors, and the causal relationship is complicated. Therefore, scholars at home and abroad have developed many computer models and software for evacuation problems. The artificial fish swarm algorithm is a new type of swarm intelligence optimization algorithm. In a water area, fish can usually swim freely or follow other individuals to find a place with more food. Therefore, the place with the largest number of fish in the water is generally the water rich in nutrients. Based on this feature, Li Xiaolei proposed an artificial fish swarm algorithm to achieve the purpose of searching and optimizing by simulating the behavior of fish foraging, tail-chasing, and swarming. Since the fish swarm algorithm was proposed, it has been widely used in robot path planning, vehicle scheduling and other issues, showing its good search ability.

When an emergency occurs, crowd evacuation efficiency has a lot to do with crowd density. When a lot of people gather at an exit, it will inevitably affect the overall evacuation. We should know from the evacuation video that the evacuation speed of individuals is different. For example, the evacuation speed of the elderly, children, and disabled groups is slow. If the crowd density at an evacuation exit is low, but most of the evacuation crowds are the above-mentioned slower action speeds. When there are three types of people, the evacuation efficiency will be greatly reduced. At the same time, if only the crowd density is used as a measure, a large number of people will gather at this exit, which will make the evacuation slower, and even cause accidents such as crowding and stampede. No relevant method has been found to take individual speed differences into account for crowd evacuation.

SUMMARY OF THE INVENTION

In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a crowd simulation evacuation method and device based on artificial fish swarm algorithm and target detection. The item of eating speed is added to the above, and the individuals who eat fast are compared to those who are slow in the evacuation, identify the individuals who are slow in the evacuation, and guide the crowd to evacuate, so as to take care of the vulnerable groups in the evacuation, and improve the evacuation efficiency at the same time. Avoid excessive congestion.

To achieve the above object, the present invention adopts the following technical solutions:

A crowd simulation evacuation method based on artificial fish swarm algorithm and target detection, comprising the following steps:

Establish a topology map according to the plan of the evacuation scene, wherein the node represents the evacuation exit, the connection line represents the connection between the two exits, and the final exit is marked;

Detect and track the individuals of the evacuated crowd according to the video data, and obtain the real-time speed of the individual evacuated crowd;

The individual real-time speed is used to represent the feeding speed in the feeding behavior of the fish school, the fish school is initialized, the food is represented by the final exit position of the scene, and the path planning for the evacuated crowd individuals is carried out based on the artificial fish swarm algorithm;

When the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is obtained.

Further, obtaining the individual real-time speed of the evacuated crowd includes:

Get the video data at the exit of the channel;

Detect pedestrian targets in the video data based on RCNN;

Pedestrian motion tracking using particle filter;

For each pedestrian, convert the actual displacement according to the position difference between the current frame and the previous frame of the pedestrian;

According to the actual displacement and the shooting frame rate of the video, the speed of the pedestrian is calculated.

Further, the detection of pedestrian targets in the video data based on RCNN includes:

Use the binary normalized gradient algorithm to select target candidate regions;

A convolutional neural network is used to extract features in the candidate area; wherein, the convolutional neural network selects a network structure comprising 5 convolutional layers and 3 fully connected layers;

The extracted features are classified by a support vector machine to identify whether the candidate region is a pedestrian target or a background.

Further, the path planning for the evacuated crowd individuals based on the artificial fish swarm algorithm includes:

Initialize the fish school, including the number, location, field of view, moving step and individual feeding speed of the artificial fish;

For each individual, the next position of the individual is determined according to the distance from the exit to be selected to the final evacuation exit, the degree of crowding and the number of individuals with a faster eating speed; wherein, pedestrians whose speed is lower than a certain threshold are used to represent the eating speed faster individuals.

Further, the determination of the next position of the individual includes:

For each individual, pre-judging the next exit selected when the individual has different behaviors at the current location, according to the distance from the selected next exit to the final evacuation exit, the degree of crowding and the number of individuals with faster eating speed Calculate the evaluation function separately; the evaluation function is:

B(vi )= _μDis (vi )+ _λδ +ωNum _s

Among them, the distance from the next exit to the final exit Y _{i represents} the food concentration at the location of the next outlet node vi; δ represents the crowding factor; Num _s represents the number of individuals who eat faster at the next outlet; μ, λ and ω represent the weight coefficients, μ+λ+ω=1; μ, λ and ω represent weight coefficients, μ+λ+ω=1;

Select the behavior with the smallest evaluation function to perform the movement of the individual.

Further, the behaviors include foraging, flocking, tail-chasing and random behaviors.

Further, the movement formulas to perform foraging, flocking, and tail-chasing behaviors are:

The move formula to perform random behavior is:

Among them, rand() is a random function with a value of 0 to 1; represents the current position of the individual, X _j represents the exit position of the target individual, Represents the position of the individual after moving; step represents the moving step; Visual is the visible range of the individual.

One or more embodiments provide a computing device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the program as claimed in claims 1-6 when executed Any one of the crowd simulation evacuation methods based on artificial fish swarm algorithm and target detection.

One or more embodiments provide a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, implements the described crowd simulation evacuation method based on artificial fish swarm algorithm and target detection.

The beneficial effects of the present invention

(1) The present invention expresses the complex environment with a topology map, succinctly expresses the connectivity problem between regions, and can intuitively reflect the evacuation process and path.

(2) The present invention takes into account the speed difference between the elderly, children, disabled groups and normal people in actual evacuation, takes pedestrian speed as an important index for selecting evacuation nodes, uses RCNN and particle filter to detect and track pedestrian motion, and can obtain real-time pedestrians. The speed reflects the real situation of exit evacuation, provides a basis for the selection of evacuation nodes, takes care of vulnerable groups during evacuation, avoids the aggravation of exit congestion, and improves evacuation efficiency.

(3) The present invention uses the fish swarm algorithm to simulate the evacuation process, and has a faster optimization speed and a good global optimization ability. By introducing the judgment of the feeding speed, when the artificial fish selects the next node, in addition to considering the crowding degree of the partner at the location, it should also consider the feeding speed of the partner. Corresponding to the crowd evacuation, when pedestrians choose the next position, considering the influence of pedestrian speed difference on the evacuation efficiency, the behavior of choosing the optimal value of the objective function shows an efficient optimization method.

Description of drawings

The accompanying drawings that form a part of the present application are used to provide further understanding of the present application, and the schematic embodiments and descriptions of the present application are used to explain the present application and do not constitute improper limitations on the present application.

Fig. 1 is the flow chart of the crowd evacuation method of the present invention;

Figure 2 is a two-dimensional plan view of an evacuation scene;

Figure 3 is a topology map generated by the connectivity of outlets;

Figure 4 is a schematic diagram of the initial random distribution of the crowd in the scene;

Figure 5 is a schematic diagram of crowd evacuation according to rules;

FIG. 6 is a schematic diagram of the end time of crowd evacuation.

Detailed ways

It should be noted that the following detailed description is exemplary and intended to provide further explanation of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

It should be noted that the terminology used herein is for the purpose of describing specific embodiments only, and is not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly dictates otherwise, the singular is intended to include the plural as well, furthermore, it is to be understood that when the terms "comprising" and/or "including" are used in this specification, it indicates that There are features, steps, operations, devices, components and/or combinations thereof.

The embodiments in this application and the features in the embodiments may be combined with each other without conflict.

The general idea proposed by the present invention: the present invention expresses the complex evacuation environment with a topological map, and expresses the connectivity of the relevant exits with the relationship between points and lines. Considering the real situation of the elderly, children, disabled groups and other individuals who move relatively slowly during the evacuation process, the present invention adds a feeding speed item on the basis of the original attributes of the artificial fish: due to individual differences in fish groups, the feeding speed is different. The density of artificial fish in the area is small, but the individual feeding speed is fast, and the food concentration in the area will decrease quickly. Therefore, before the artificial fish performs the swarming behavior, it should be pre-judged whether there are individuals who eat too fast in the fish group, so as to avoid overwork. No use. The individual who eats fast is likened to the individual who moves slowly during evacuation, and the evacuation speed of the individual is used as an important indicator to be measured in the evacuation, which avoids the exit congestion caused by ignoring the difference of individual evacuation speed only by using density as the measure, and improves the evacuation efficiency. . Use RCNN to identify pedestrians in motion, and further calculate their speed, count the number of individuals moving slowly in a certain area, provide a basis for the selection of pedestrian evacuation exits, reflect the real situation of evacuation, and avoid secondary accidents.

Topological map is an important representation method of the environment in robotics. It represents the indoor environment as a topological structure map with nodes and related connecting lines, where the nodes represent important locations in the environment (corners, doors, elevators, stairs, etc.) , and the edge represents the connection relationship between nodes, such as corridors. Topological map is a simplified and regularized representation of the actual environment. It uses nodes and lines to establish positional relationships, which has low complexity and can simply and effectively represent the problem of exit path selection in crowd evacuation.

RCNN (Regions with CNN features) is a target detection method, which is a milestone in the application of convolutional neural network methods to target detection problems. The Region Proposal method realizes the transformation of the target detection problem. It is accurate and efficient, and can meet the needs of real-time problems.

Example 1

This embodiment discloses a crowd simulation evacuation method based on artificial fish swarm algorithm and target detection, as shown in FIG. 1 , including the following steps:

Step 1: Obtain the connectivity between areas according to the plan of the evacuation scene, and establish a topology map, in which the nodes represent the evacuation exits, the connecting lines represent the connectivity between the two exits, and the final exits are marked.

Obtain a two-dimensional plan of the evacuation environment, and build a topology map according to the connectivity of the area. The evacuation exits are numbered vi ( _i =1,2,...n), the nodes represent the evacuation exits, and the connection lines represent the connectivity between the two exits. As shown in Figure 2, the nodes marked with ⑤ and ⑥ represent the final exit.

Step 2: Detect and track pedestrians according to the video data to obtain the real-time speed of pedestrians.

Use RCNN and particle filter to detect and track pedestrians evacuated from an exit, obtain the real-time speed of pedestrians, calculate the number of slow-moving individuals gathered at an exit, and provide a basis for pedestrians to choose exits. Screen out s individuals at a certain exit, and count the number of them.

Specifically, a CCD camera is installed at each exit of the evacuation scene to detect and track pedestrians in an all-round way. No matter which exit a pedestrian walks from, their real-time speed can be measured. After acquiring the video data, first use the convolutional neural network to extract pedestrian features, track them, and measure the pedestrian speed. Specific steps include:

RCNN is used to detect pedestrian targets in surveillance video, which is described as follows:

Use the binary normalized gradient algorithm BING to select possible targets and obtain candidate regions;

Feature extraction is performed within the candidate region using a convolutional neural network algorithm (CNN). The Alex net network structure containing 5 convolutional layers and 3 fully connected layers can be selected. First, the input vector x and the filter weights are convolved, and the calculation formula is:

f=g(∑ _i∈M x _i ×w+b) (1)

Increase the bias b, then use the ReLU activation function g=max(0,x), and finally perform maximum pooling and input to the next layer. The last fully connected layer outputs a 4096-dimensional feature vector. The features extracted by CNN are finally classified by SVM to identify candidate regions as pedestrian targets or backgrounds.

Particle filters are then used for predictive tracking of pedestrian movements, described as follows:

Suppose Y={y ¹ , y ² ,...y ^t } represents the observed variable sequence, X={x ¹ , x ² ..., x ^t } represents the tracked vector sequence, y ^t and x ^t represent the observed variables at time t respectively and the state variable of the tracked target, the weight is w ^t . Then the posterior probability p is

Comparing the similarity between the current particle and the previous target can get the position of the target in the next frame. Particles are sampled from the importance distribution q(x ^t |x ^1:1-t , y ^1:t ). The particle position with the highest similarity is determined by the weight, and the weight is updated according to

After the weight is updated, if it is less than a threshold, the particles need to be resampled to eliminate particles with too low weights.

After obtaining the coordinate position of the pedestrian in the next frame, compare it with the coordinates of the previous frame to obtain the pedestrian displacement l in the image. According to the corresponding relationship between the pedestrian displacement l and the actual displacement in the image, the actual displacement l′ is obtained. Since the shooting frame rate of the camera is constant, the pedestrian speed v can be calculated.

Through the above process, the speed of the pedestrian to be evacuated from a certain exit can be measured. When the pedestrian's speed v is lower than 1m/s, it can be considered that the pedestrian's evacuation speed is relatively slow, which is defined as s, and the area s is recorded. Number Num _s . Set the maximum number T of s without affecting the evacuation of an exit.

Step 3: Divide the pedestrians with faster and slower speeds according to a certain threshold, respectively corresponding to the slower and faster individuals in the feeding behavior of the fish;

The feeding behavior of the fish is defined on the basis of the original attributes of the fish, and divided into fast-feeding individuals and general individuals according to the feeding speed. Correspondingly, for the individuals with slow evacuation speed and ordinary individuals during the evacuation process, the individuals with slow evacuation speed are placed in a certain The number of exits is counted to guide pedestrians to evacuate and avoid increased congestion. The feeding speed of artificial fish is divided into two types: normal speed and fast speed, which correspond to the average individual and the slow-moving individual during evacuation, respectively. The individuals who eat too fast (individuals who move slowly) are denoted as s, and their number is denoted as Num _s .

Statistics show that in an emergency, the walking speed of ordinary people is 1.25m/s, the walking speed of the elderly or children is 0.65m/s, and the speed of disabled people who need help is 0.57m/s. In the case of , take 1m/s as the threshold, and the individual whose walking speed is below the threshold is denoted as s. For evacuation efficiency and safety considerations, it is necessary to screen s individuals and count their number to reflect the real evacuation situation and evacuation capability of the exit, provide a basis for the correct choice of pedestrians, and avoid aggravation of exit congestion. Screen out s individuals at a certain exit, and count the number of them.

Step 4: Initialize the fish swarm, represent the food with the final exit position of the scene, and plan the path of the evacuated crowd based on the artificial fish swarm algorithm.

Simulation of crowd evacuation using artificial fish swarms searching for food. The basis of artificial fish judging the quality of a node includes food concentration, crowding factor and the number of fast-eating individuals. There are three behaviors of artificial fish: foraging, flocking, and tail-chasing, from which the behavior that minimizes the value of the objective function is selected.

Initialize the fish school, including the number, location, field of view, moving step, and feeding speed of the artificial fish. The feeding speed of the individual is divided into fast and normal. Guide pedestrians to evacuate according to the fish swarm algorithm. The vector X=(x ₁ , x ₂ ,...,x _n ) represents the state of the artificial fish individual, where x _i (i=1, 2,...n) is the variable to be searched for optimization; Y is the food concentration; d _{i ,j} =||X _i -X _j || is the distance between individuals; Visual is the visible range of individuals; step is the moving step size; δ(0<δ<1) is the crowding factor; Trynum is the maximum number of attempts .

The crowd evacuation is simulated by the fish swarm searching for food. In the fish swarm algorithm, the smaller the distance between the artificial fish and the food, the greater the food concentration. The food position can represent the final exit position in the crowd evacuation. An exit node v _i and the final evacuation exit The distance Dis(v _i ) can be expressed as The selection of the exit node is determined by the distance Dis(vi ₎ from the exit node v _i to the final evacuation exit, the congestion factor δ of the node and the number of s individuals at the node Nums. The evaluation function can be expressed as

B(v _i )=μDis(vi ₎ +λδ+ωNum _s , where μ+λ+ω=1

Simulate the next node selected by individuals when foraging, swarming, tail-chasing and random behaviors occur at the current position, respectively. The evaluation function is used to judge the pros and cons of the next node, and the artificial fish chooses the behavior with the smallest evaluation function to execute.

The behavior of the artificial fish is expressed as follows:

Foraging: people randomly choose an evacuation path in the process of panic, which is similar to the foraging behavior of fish, and the evaluation function is only determined by the distance between the next node and the final exit Dis( _vi ). Assuming that pedestrian x _i is at the v _t node at this time, and randomly selects the node v _p within its field of view, if the food concentration of the v _p node Y _P > Y _t , then v _p can be selected as the next node, the evaluation function Denote it as B(v _i ) _forage ; otherwise select other nodes. After repeating Trynum times as above, the artificial fish performs random behavior. The moving formula is:

rand() is a random function with a value of 0 to 1.

Swarming: During the evacuation process, people tend to have a herd mentality, similar to the swarming behavior of fish. In the swarming behavior, the target that the artificial fish wants to choose is the exit position of the central fish in the field of vision. Assuming that the pedestrian _xi is at the v _t node at this time, obtain the gathering center node v _c of the partners within its field of vision and the number of partners n _f at the node, if Prove that the central location has sufficient food and the density of partners is not very high. At this time, the relationship between the number of s individuals in the partner and the threshold T is judged. If Num _s < T, v _c can be selected as the next node, and the evaluation function is recorded as B(v _i ) _bunching . The movement formula is expressed as follows:

Otherwise perform other actions.

Rear chasing: Pedestrians in the evacuation process tend to follow others to evacuate due to their blindness. Similar to rear chasing in the fish swarm algorithm, the target to be selected in the rear chasing behavior is the exit position of the fish with the best state in the field of vision. Assuming that pedestrian x _i is at node v _t at this time, obtain the node v _p of the partner with the optimal state in its field of vision and the number of its surrounding partners n _f , if Prove that there is more food in this state and the density of partners is not very large. Further judge the relationship between the number of s individuals in the partner and the threshold T. If Num _s < T, then v _p can be selected as the next node, and the evaluation function is recorded as B ( _vi ) _pursue .

The moving formula is:

Otherwise choose other behavior.

Random: Random behavior is a default behavior of foraging behavior. The artificial fish randomly selects a state or a partner, and the evaluation function is denoted as B( _vi ) _random . The moving formula is:

After pre-judging the next behavior, the artificial fish selects the behavior that makes the evaluation function take the minimum value, that is,

min(B(v _i ) _forage , B(vi _i ) _bunching , B(vi _i ) _pursue , B(vi _i ) _random ).

Step 5: When the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is derived.

That is, it ends when the number of people coming out of the final exit is equal to the total number of people, as shown in FIG. 6 .

Embodiment 2

The purpose of this embodiment is to provide a computing device.

A computing device, comprising a memory, a processor and a computer program stored in the memory and running on the processor, the processor implements the following steps when executing the program, including:

Establish a topology map according to the plan of the evacuation scene, wherein the node represents the evacuation exit, the connection line represents the connection between the two exits, and the final exit is marked;

Detect and track the individuals of the evacuated crowd according to the video data, and obtain the real-time speed of the individual evacuated crowd;

The individual real-time speed is used to represent the feeding speed in the feeding behavior of the fish school, the fish school is initialized, the food is represented by the final exit position of the scene, and the path planning for the evacuated crowd individuals is carried out based on the artificial fish swarm algorithm;

When the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is obtained.

Embodiment 3

The purpose of this embodiment is to provide a computer-readable storage medium.

A computer-readable storage medium on which a computer program is stored, the program executes the following steps when executed by a processor:

Establish a topology map according to the plan of the evacuation scene, wherein the node represents the evacuation exit, the connection line represents the connection between the two exits, and the final exit is marked;

Detect and track the individuals of the evacuated crowd according to the video data, and obtain the real-time speed of the individual evacuated crowd;

The individual real-time speed is used to represent the feeding speed in the feeding behavior of the fish school, the fish school is initialized, the food is represented by the final exit position of the scene, and the path planning for the evacuated crowd individuals is carried out based on the artificial fish swarm algorithm;

When the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is obtained.

The steps involved in Embodiments 2 and 3 above correspond to Embodiment 1 of the method, and the specific implementation can refer to the relevant description part of Embodiment 1. The term "computer-readable storage medium" should be understood to include a single medium or multiple media including one or more sets of instructions; it should also be understood to include any medium capable of storing, encoding or carrying for use by a processor The executed instruction set causes the processor to perform any of the methods of the present invention.

The beneficial effects of the present invention

(1) The present invention expresses the complex environment with a topology map, succinctly expresses the connectivity problem between regions, and can intuitively reflect the evacuation process and path.

(2) The present invention takes into account the speed difference between the elderly, children, disabled groups and normal people in actual evacuation, takes pedestrian speed as an important index for selecting evacuation nodes, uses RCNN and particle filter to detect and track pedestrian motion, and can obtain real-time pedestrians. The speed reflects the real situation of exit evacuation, provides a basis for the selection of evacuation nodes, takes care of vulnerable groups during evacuation, avoids the aggravation of exit congestion, and improves evacuation efficiency.

(3) The present invention uses the fish swarm algorithm to simulate the evacuation process, and has a faster optimization speed and a good global optimization ability. By introducing the judgment of the feeding speed, when the artificial fish selects the next node, in addition to considering the crowding degree of the partner at the location, it should also consider the feeding speed of the partner. Corresponding to the crowd evacuation, when pedestrians choose the next position, considering the influence of pedestrian speed difference on the evacuation efficiency, the behavior of choosing the optimal value of the objective function shows an efficient optimization method.

Those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by the computing device, so that they can be stored in a storage device. The device is executed by a computing device, or they are separately fabricated into individual integrated circuit modules, or multiple modules or steps in them are fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

The above descriptions are only preferred embodiments of the present application, and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application shall be included within the protection scope of this application.

Although the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, they do not limit the scope of protection of the present invention. Those skilled in the art should understand that on the basis of the technical solutions of the present invention, those skilled in the art do not need to pay creative work. Various modifications or variations that can be made are still within the protection scope of the present invention.

Claims (7)

1. a crowd simulation evacuation method based on artificial fish swarm algorithm and target detection, is characterized in that, comprises the following steps: Establish a topology map according to the plan of the evacuation scene, wherein the node represents the evacuation exit, the connection line represents the connection between the two exits, and the final exit is marked; Detect and track the individuals of the evacuated crowd according to the video data, and obtain the real-time speed of the individual evacuated crowd; The individual real-time speed is used to represent the feeding speed in the feeding behavior of the fish school, the fish school is initialized, the food is represented by the final exit position of the scene, and the path planning for the evacuated crowd individuals is carried out based on the artificial fish swarm algorithm; When the number of evacuees at the final exit is equal to the total number of people, the evacuation process ends, and the evacuation path is obtained; The path planning for individual evacuated crowds based on the artificial fish swarm algorithm includes: Initialize the fish school, including the number, location, field of view, moving step and individual feeding speed of the artificial fish; For each individual, the next position of the individual is determined according to the distance from the selectable exit to the final exit, the crowding degree of the selectable exit location, and the number of individuals with faster eating speed; among them, pedestrians whose speed is lower than a certain threshold are used Indicates that the individual who eats faster; The determination of the next location of the individual includes: For each individual, pre-judging the next exit selected when the individual has different behaviors at the current location, according to the distance from the selected next exit to the final evacuation exit, the degree of crowding and the number of individuals with faster eating speed Calculate the evaluation function separately; the evaluation function is: B(vi )= _μDis (vi )+ _λδ +ωNum _s Among them, the distance from the next exit to the final exit Y _{i represents} the food concentration at the location of the next outlet vi; δ represents the crowding factor; Num _s represents the number of individuals who eat faster at the next outlet; μ, λ and ω represent the weight coefficient, μ+ λ+ω=1; Select the behavior with the smallest evaluation function to perform the movement of the individual. 2. a kind of crowd simulation evacuation method based on artificial fish swarm algorithm and target detection as claimed in claim 1 is characterized in that, obtaining the individual real-time speed of evacuation crowd comprises: Get the video data at the exit of the channel; Detect pedestrian targets in the video data based on RCNN; Pedestrian motion tracking using particle filter; For each pedestrian, convert the actual displacement according to the position difference between the current frame and the previous frame of the pedestrian; According to the actual displacement and the shooting frame rate of the video, the speed of the pedestrian is calculated. 3. a kind of crowd simulation evacuation method based on artificial fish swarm algorithm and target detection as claimed in claim 2, is characterized in that, described based on RCNN detects the pedestrian target in described video data comprises: Use the binary normalized gradient algorithm to select target candidate regions; A convolutional neural network is used to extract features in the candidate area; wherein, the convolutional neural network selects a network structure comprising 5 convolutional layers and 3 fully connected layers; The extracted features are classified by a support vector machine to identify whether the candidate region is a pedestrian target or a background. 4 . A crowd simulation evacuation method based on artificial fish swarm algorithm and target detection according to claim 1 , wherein the behaviors include foraging, swarming, tail-chasing and random behaviors. 5 . 5. a kind of crowd simulation evacuation method based on artificial fish swarm algorithm and target detection as claimed in claim 4, is characterized in that, the movement formula that occurs foraging, swarming, rear-chasing behavior is: The move formula to perform random behavior is: Among them, rand() is a random function with a value of 0 to 1; represents the current position of the individual, X _j represents the exit position of the target individual, Represents the position of the individual after moving; step represents the moving step; Visual is the visible range of the individual. 6. A computing device comprising a memory, a processor and a computer program stored on the memory and running on the processor, wherein the processor implements any one of claims 1-5 when the processor executes the program The crowd simulation evacuation method based on artificial fish swarm algorithm and target detection described in item. 7. A computer-readable storage medium on which a computer program is stored, characterized in that, when the program is executed by the processor, the artificial fish swarm algorithm and target detection based on any one of claims 1-5 are realized. Crowd simulation evacuation method.