Universal automatic aiming method and device for all game engines
By adopting lightweight YOLO model and Kaman filter prediction technology, combined with GPU acceleration and hardware-level input simulation, compatibility and efficient automatic targeting of all game engines are achieved, solving the shortcomings of the existing technology in terms of compatibility, security and real-time.
Patent Information
- Application Number
- CN202510419113.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-03
- Publication Date
- 2025-05-30
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing self-targeting technology has shortcomings in compatibility, security and real-time, especially the memory read-based scheme is easily detected by anti-cheating systems, and the computer vision-based scheme is computationally expensive and affects the user experience.
The lightweight YOLO model is used for real-time target detection, and combined with GPU acceleration and multi-threading optimization, the priority of the shooting target is dynamically determined, the target position and speed are predicted based on Kaman filtering, and a mouse trajectory that conforms to human behavior characteristics is generated, and the natural mouse movement trajectory is simulated through hardware-level input to shoot.
It realizes smooth automatic aiming capabilities in high frame rate environments, improves the hit rate of dynamic targets, reduces aiming deviations, and avoids the risk of account bans caused by traditional self-targeting solutions due to modifying memory or calling game APIs.
Smart Images
Figure CN120053993A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure belongs to the field of computer technology, and particularly relates to an automatic aiming method and device that are common to all game engines. Background Art
[0002] The auto-aim (Aimbot) technology is a software in video games, especially in shooting games, which is used to assist players in automatically or manually aiming at enemies, thereby greatly reducing the difficulty of aiming and enabling players to easily hit enemies. To ensure game fairness, games often detect and restrict players from using auto-aim technology in the game.
[0003] Currently, there are various auto-aim technology solutions on the market, but they all have deficiencies in terms of compatibility, security, real-time performance, etc. For example, the solution based on pixel color detection can aim by reading a specific color on the screen (such as the color of the enemy character), which is simple to implement but easily interfered with; the solution based on memory reading can directly obtain the coordinates of players and enemies by scanning the game memory and then perform automatic aiming. This method has high accuracy but is easily detected by anti-cheat systems; while the solution based on computer vision (CV) combined with machine learning analyzes the game screen by using a target detection model (such as YOLO, OpenCVDNN) and combines mouse / touch simulation to perform auto-aim. This method is applicable to almost all games, but has a large amount of computation. Summary of the Invention
[0004] In view of the above problems, various embodiments of the present disclosure propose an automatic aiming solution that is common to all game engines.
[0005] The first aspect of the embodiments of the present disclosure provides an automatic aiming method that is common to all game engines, including:
[0006] Obtain a game picture, and input the game picture into a target recognition model to identify a shooting target and return the bounding box coordinates, where the target recognition model is trained and generated by using a data set composed of game screenshots obtained under various conditions to enhance a lightweight YOLO model;
[0007] Dynamically determine the priority of the shooting target based on the target distance, threat level, and field of view angle, and predict the position and speed of high-priority targets based on Kalman filtering;
[0008] Generate a mouse trajectory that conforms to human behavior characteristics based on the position and speed of the high-priority target, and complete the shooting action at the target position by using hardware-level input to simulate the natural mouse movement trajectory.
[0009] In some embodiments of the present disclosure, the obtaining of the game picture includes:
[0010] Obtain game pictures and decode them multi-threadedly based on a screen capture tool.
[0011] In some embodiments of the present disclosure, after obtaining the game pictures, it further includes:
[0012] Preprocess the game pictures using GPU acceleration, where the preprocessing at least includes noise reduction and sharpening.
[0013] In some embodiments of the present disclosure, the lightweight processing of the YOLO model includes:
[0014] Crop some deep convolutional layers and use depthwise separable convolutions instead of standard convolutions; and / or
[0015] Reduce the number of anchor boxes for the game scene; and / or
[0016] Adopt INT8 quantization to reduce the video memory occupancy; and / or
[0017] Use the teacher model to guide the training of the small model to prune redundant weights and improve the detection accuracy.
[0018] In some embodiments of the present disclosure, obtaining game screenshots under multiple conditions includes:
[0019] Take screenshots under different maps, weather, and lighting conditions to cover various target postures;
[0020] Use annotation tools to annotate the targets in the screenshots.
[0021] In some embodiments of the present disclosure, the enhancement processing of the game screenshots includes:
[0022] Adjust the brightness and contrast of the game screenshots to adapt to different game environments; and / or
[0023] Perform rotation, cropping, and scaling on the game screenshots to improve the robustness of the model; and / or
[0024] Perform color transformation on the game screenshots to prevent the model from overfitting to a specific game color style; and / or
[0025] Mix the game screenshots to enhance the target diversity and improve the generalization ability.
[0026] In some embodiments of the present disclosure, it is characterized in that the dynamic determination of the priority of the shooting target based on the target distance, threat level, and field of view angle includes:
[0027] Dynamically determine the priority of the shooting target based on the following formula:
[0028]
[0029] where P is the priority of the shooting target, d, T, and θ are the target distance, threat level, and field of view angle of the shooting target respectively, and w 1 , w 2 , w 3 are adjustable weight coefficients, and α, β, and γ are adjustment parameters used to adjust the influence of different factors on the priority.
[0030] In some embodiments of the present disclosure, the predicting the position and velocity of a high-priority target based on Kalman filtering includes:
[0031] Determining the current position and velocity of the high-priority target based on Kalman filtering;
[0032] Predicting the position and velocity of the high-priority target in the next frame based on the following formula:
[0033] v k+1 = v k + aΔt
[0034] where x k and x k+1 are the current position and the position in the next frame of the high-priority target respectively, v k and v k+1 are the current velocity and the velocity in the next frame of the high-priority target respectively; Δt is the time interval between two frames, and a is the acceleration.
[0035] In some embodiments of the present disclosure, the mouse trajectory is a Bezier curve.
[0036] A second aspect of the embodiments of the present disclosure provides an automatic aiming device that is common to all game engines, including:
[0037] An identification module for acquiring a game picture, inputting the game picture into a target recognition model to identify a shooting target and returning the bounding box coordinates, where the target recognition model is trained and generated using a data set composed of game screenshots obtained under various conditions after being enhanced for a lightweight YOLO model;
[0038] A prediction module for dynamically determining the priority of the shooting target based on the target distance, threat level, and field of view angle, and predicting the position and velocity of the high-priority target based on Kalman filtering;
[0039] A shooting module for generating a mouse trajectory that conforms to human behavior characteristics based on the position and velocity of the high-priority target, and completing the shooting action on the target position by simulating the natural mouse movement trajectory using hardware-level input.
[0040] In summary, the automatic aiming methods and devices provided by the embodiments of the present disclosure, which are common to all game engines, perform low-latency real-time target detection by using a lightweight YOLO model and combine GPU acceleration and multi-thread optimization to improve the processing speed, ensuring smooth automatic aiming ability in a high-frame-rate environment; and optimize the aiming object by combining intelligent target screening and prediction with the target priority strategy and adopt a target movement prediction algorithm to improve the hit rate of dynamic targets and reduce aiming deviation; finally, adopt hardware-level input simulation technology to simulate the natural mouse movement trajectory, avoid anti-cheat detection, and avoid the risk of being banned due to modifying the memory or calling the game API in traditional aimbot solutions. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] The features and advantages of the present disclosure will be more clearly understood by referring to the accompanying drawings. The drawings are schematic and should not be construed as imposing any limitation on the present disclosure. In the drawings:
[0042] Figure 1 is a flowchart of an automatic aiming method common to all game engines according to some embodiments of the present disclosure;
[0043] Figure 2 is a schematic diagram of an automatic aiming device common to all game engines according to some embodiments of the present disclosure;
[0044] Figure 3 is a schematic diagram of an automatic aiming device common to all game engines according to some embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0045] In the following detailed description, many specific details of the present disclosure are set forth by way of example in order to provide a thorough understanding of the relevant disclosure. However, it will be apparent to those of ordinary skill in the art that the present disclosure may be practiced without these details. It should be understood that the terms "system", "device", "unit" and / or "module" used in the present disclosure are a way of distinguishing different components, elements, parts or assemblies at different levels in a sequential arrangement. However, if other expressions can achieve the same purpose, these terms may be replaced by other expressions.
[0046] It should be understood that when a device, unit or module is referred to as being "on", "connected to" or "coupled to" another device, unit or module, it may be directly on, connected to or coupled to or communicate with the other device, unit or module, or there may be intermediate devices, units or modules, unless the context clearly indicates otherwise. For example, the term "and / or" used in the present disclosure includes any and all combinations of one or more of the associated listed items.
[0047] The terms used in this disclosure are only for describing specific embodiments and do not limit the scope of this disclosure. As shown in the specification and claims of this disclosure, unless the context clearly indicates otherwise, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified features, wholes, steps, operations, elements, and / or components, and such expressions do not constitute an exclusive list, and other features, wholes, steps, operations, elements, and / or components may also be included.
[0048] Referring to the following description and the accompanying drawings, these or other features and characteristics of the present disclosure, the operating methods, the functions of the related elements of the structure, the combination of parts, and the economy of manufacture can be better understood, where the description and the drawings form a part of the specification. However, it can be clearly understood that the drawings are only for the purpose of illustration and description and are not intended to limit the protection scope of the present disclosure. It can be understood that the drawings are not drawn to scale.
[0049] A variety of structure diagrams are used in this disclosure to illustrate various deformations according to the embodiments of the present disclosure. It should be understood that the structures before or below are not used to limit the present disclosure. The protection scope of the present disclosure is subject to the claims.
[0050] Aimbot is a technology used to automatically aim at targets in video games, which is widely used in first-person shooter (FPS), third-person shooter (TPS), and some MOBA games to assist players in automatically or manually aiming at enemies, thus greatly reducing the difficulty of aiming and enabling players to easily hit enemies. Its main technical core lies in target detection, crosshair positioning, input simulation, etc.
[0051] Currently, there are various aimbot technology solutions on the market, but they all have deficiencies in terms of compatibility, security, real-time performance, etc.
[0052] A common method is the aimbot technology based on game memory reading. This solution directly extracts the target coordinate data from the memory of the game process and uses mathematical calculations to automatically adjust the position of the crosshair. The advantage of this method is high accuracy and low computational complexity, and it can complete aiming quickly, but there is a serious risk of anti-cheat detection. Modern anti-cheat systems use means such as memory integrity verification and kernel-level process monitoring to quickly detect and ban memory modification type cheats, making this solution have great problems in terms of security.
[0053] Another relatively new auto-aiming technology is based on computer vision (CV) + object detection. This solution identifies targets by analyzing the game screen (instead of reading memory) and simulates input for automatic aiming. The biggest advantage of this method is that it bypasses the memory detection mechanism of anti-cheat systems, reduces the risk of account suspension, and can be applied to almost all game engines (Unity, Unreal, Cocos, IL2CPP, Mono, etc.). However, the main challenge of this solution is the large computational cost. If not optimized properly, it may cause auto-aiming latency and affect the user experience. In addition, the accuracy of object detection also depends on the training quality of the model and may be affected by environmental factors (such as game special effects, resolution changes, character occlusion, etc.).
[0054] In addition, some auto-aiming solutions use deep learning models (such as YOLO, MobileNet, etc.) for object detection and combine reinforcement learning for optimization, making the auto-aiming path more user-friendly. However, the training cost of deep learning models is high, and they have high requirements for hardware performance, making it difficult to achieve efficient operation on mobile devices.
[0055] To solve the above problems, the present disclosure proposes a YOLO auto-aiming technology method based on a lightweight YOLO model, which is applicable to all game engines. In some embodiments, the flowchart of the automatic aiming common to all game engines is shown in FIG. 1 and specifically includes the following steps:
[0056] S110, obtain a game picture, input the game picture into an object recognition model to recognize a shooting target and return the bounding box coordinates, wherein the object recognition model is trained and generated by using a dataset composed of game screenshots enhanced under various conditions for a lightweight YOLO model.
[0057] First, select a suitable screen capture method (DXGI is available for Windows, and X11 is available for Linux). Simultaneously obtain images and decode them through multi-threading. And perform preprocessing on the game image, such as noise reduction and sharpening, by using GPU acceleration (CUDA / OpenCL) to improve the detection accuracy. GPU acceleration can reduce the CPU load and reduce the capture latency.
[0058] Then, input the game image into the object recognition model to recognize the shooting target and return the bounding box coordinates.
[0059] In some embodiments of the present disclosure, the object recognition model is trained from a lightweight YOLO model using a custom dataset. It includes:
[0060] 1. Reduce the depth of the convolutional layer: Crop some deep convolutional layers to reduce the computational amount and maintain the core feature extraction ability.
[0061] 2. Reduce the number of Anchor Boxes: Optimize Anchor Boxes for the game scenario, reduce redundant calculations, and improve the efficiency of target matching.
[0062] 3. Replace standard convolution with depthwise separable convolution: Improve the inference speed.
[0063] 4. Quantize the model: Adopt INT8 quantization to reduce the VRAM occupancy while maintaining high accuracy.
[0064] 5. Pruning and Knowledge Distillation: Prune redundant weights and use the teacher model to guide the training of the small model to improve the detection accuracy.
[0065] The construction method of the custom dataset is as follows:
[0066] 1. Data collection:
[0067] Game screenshots: Take screenshots under different maps, weather, and lighting conditions, covering various enemy postures.
[0068] Automatic annotation: Use a semi-automatic annotation tool (such as LabelImg) to annotate the enemies in the screenshots.
[0069] 2. Data augmentation:
[0070] Brightness and contrast adjustment: Adapt to different game environments.
[0071] Rotation, cropping, and scaling: Improve the robustness of the model.
[0072] Color transformation: Prevent the model from overfitting to a specific game color style.
[0073] MixUp and CutMix: Enhance the diversity of targets and improve the generalization ability.
[0074] 3. Dataset format:
[0075] Adopt the COCO or VOC format to facilitate compatibility with YOLO training.
[0076] Generate the training set, validation set, and test set to optimize the model performance.
[0077] Finally, the optimization of the present disclosure reduces the YOLO model size by 40%+ and improves the detection speed by 2 - 3 times, while still maintaining high recognition accuracy, which is applicable to real-time target detection in games.
[0078] S120, Dynamically determine the priority of the shooting target based on the target distance, threat level, and field of view angle, and predict the position and speed of the high-priority target based on the Kalman filter.
[0079] The present disclosure calculates the target priority based on a dynamic priority algorithm (based on enemy distance, threat level, and field of view angle), and then combines Kalman filtering for movement prediction to improve the hit rate and avoid misaiming.
[0080] Specifically, in some embodiments of the present disclosure, the dynamic priority algorithm is:
[0081]
[0082] Where P is the priority of the shooting target, d, T, and θ are the target distance, threat level, and field of view angle of the shooting target respectively, w 1 , w 2 , w 3 are adjustable weight coefficients, and α, β, and γ are adjustment parameters used to adjust the influence of different factors on the priority.
[0083] Some embodiments of the present disclosure are based on the exponential decay term e -αd to ensure that the priority of distant targets is reduced while retaining the ability for dynamic adjustment; at the same time, the non - linear threat weighting T β gives a greater weight to high - threat targets, thereby non - linearly enhancing the priority; the denominator is normalized with (1 + γT) to prevent numerical overflow and improve stability.
[0084] Then, based on Kalman filtering, the position and velocity of high - priority targets are predicted:
[0085] First, based on Kalman filtering, the current position and velocity of the high - priority target are determined;
[0086] Then, based on the following formula, the position and velocity of the high - priority target in the next frame are predicted:
[0087] v k+1 = v k + aΔt
[0088] Where x k and x k+1 are the current position and the position in the next frame of the high - priority target respectively, v k and v k+1 are the current velocity and the velocity in the next frame of the high - priority target respectively; Δt is the time interval between two frames, and a is the acceleration.
[0089] Kalman filtering can effectively predict the position and velocity of the target in the next frame, making the self - aiming more accurate.
[0090] S130. Generate a mouse trajectory that conforms to human behavior characteristics based on the position and speed of the high-priority target, and complete the shooting action at the target position by simulating the natural mouse movement trajectory using hardware-level input.
[0091] Some embodiments of the present disclosure use hardware-level input (such as Arduino Leonardo, HID USB devices) to simulate the natural mouse movement trajectory, avoid directly operating using APIs such as SendInput, and prevent anti-cheat detection. Specifically:
[0092] Calculate the target position and generate a mouse trajectory that conforms to human behavior characteristics (such as a Bezier curve).
[0093] Simulate mouse input through the HID device, avoid calling SetCursorPos(), and improve concealment.
[0094] Introduce random jitter to simulate the real player's hand shaking and make the trajectory more natural.
[0095] Figure 2 It is a schematic diagram of an automatic aiming device that is common to all game engines shown in some embodiments of the present disclosure. As Figure 2 shown, the automatic aiming device 200 that is common to all game engines includes an identification module 210, a prediction module 220, and a shooting module 230. Among them:
[0096] The identification module 210 is used to obtain a game picture, input the game picture into a target recognition model to identify the shooting target and return the bounding box coordinates. Among them, the target recognition model is trained and generated using a data set composed of game screenshots obtained under various conditions after being light-weight processed for the YOLO model;
[0097] The prediction module 220 is used to dynamically determine the priority of the shooting target based on the target distance, threat level, and field of view angle, and predict the position and speed of the high-priority target based on the Kalman filter;
[0098] The shooting module 230 is used to generate a mouse trajectory that conforms to human behavior characteristics based on the position and speed of the high-priority target, and complete the shooting action at the target position by simulating the natural mouse movement trajectory using hardware-level input.
[0099] Some embodiments of the present disclosure disclose a computer program product, including computer programs / instructions, which when executed by a processor implement Figure 1 the method of automatic aiming that is common to all game engines described in S110 - S130 above.
[0100] Figure 3Schematic diagram of an electronic device according to some embodiments of the present disclosure. As Figure 3 shown, the electronic device 300 includes a memory 320 and a processor 310. The memory 320 is used to store a computer program. The processor 310 is used to implement Figure 1 the method of automatic aiming common to all game engines described in S110 - S130 in
[0101] Some embodiments of the present disclosure disclose a storage medium on which computer - executable instructions are stored. When the computer - executable instructions are executed by a computing device, they can be used to implement Figure 1 the method of automatic aiming common to all game engines described in S110 - S130 in
[0102] In summary, for the method and device of automatic aiming common to all game engines provided in the embodiments of the present disclosure, by using a lightweight YOLO model for low - latency real - time target detection and combining GPU acceleration and multi - thread optimization to improve the processing speed, it ensures the ability to maintain smooth automatic aiming in a high - frame - rate environment. And by combining intelligent target screening and prediction with a target priority strategy to optimize the aiming object and using a target movement prediction algorithm, it improves the hit rate of dynamic targets and reduces aiming deviation. Finally, by adopting hardware - level input simulation technology to simulate the natural mouse movement trajectory, it avoids anti - cheating detection and circumvents the risk of being banned due to modifying memory or calling game APIs in traditional auto - aiming solutions.
[0103] Although the subject matter described herein is provided in the general context of execution with an operating system and application programs on a computer system, those skilled in the art will recognize that other implementations can also be performed in combination with other types of program modules. Generally, program modules include routines, programs, components, data structures, and other types of structures that perform specific tasks or implement specific abstract data types. Those skilled in the art can understand that the subject matter described herein can be practiced using other computer system configurations, including handheld devices, multi - processor systems, microprocessor - based or programmable consumer electronics, minicomputers, mainframe computers, etc., and can also be used in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in both local and remote memory storage devices.
[0104] Those of ordinary skill in the art can realize that the units and method steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of this disclosure.
[0105] It should be understood that the above specific embodiments of the present disclosure are only used for illustrative explanation of the principles of the present disclosure and do not constitute a limitation to the present disclosure. Therefore, any modifications, equivalent replacements, improvements, etc. made without departing from the spirit and scope of the present disclosure shall be included within the protection scope of the present disclosure. In addition, the appended claims of the present disclosure are intended to cover all changes and modification examples falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.
Claims
1. An automatic aiming method common to all game engines, characterized in that: include: Obtaining a game image, and inputting the game image into a target recognition model to identify a shooting target and return bounding box coordinates, wherein the target recognition model is generated by training a lightweight YOLO model with a data set consisting of enhanced game screenshots obtained under various conditions; Dynamically determine the priority of the shooting target based on target distance, threat level and field of view angle, and predict the position and speed of high priority targets based on Kalman filtering; A mouse trajectory that conforms to human behavior characteristics is generated based on the position and speed of the high-priority target, and a natural mouse movement trajectory is simulated using hardware-level input to complete a shooting action at the target position.
2. The method according to claim 1, characterized in that The obtaining of game pictures comprises: Based on the screen capture tool, game images are obtained and decoded in multi-threaded manner.
3. The method according to claim 1, characterized in that After obtaining the game picture, the method further includes: The game image is preprocessed using GPU acceleration, wherein the preprocessing includes at least noise reduction and sharpening.
4. The method according to claim 1, characterized in that: The lightweight processing of the YOLO model includes: Crop some deep convolutional layers and use depthwise separable convolutions instead of standard convolutions; and / or Reduce the number of anchor boxes for gaming scenarios; and / or Use INT8 quantization to reduce video memory usage; and / or The teacher model is used to guide the training of the small model to remove redundant weights and improve detection accuracy.
5. The method according to claim 1, characterized in that: Capture game screenshots under various conditions including: Take screenshots under different maps, weather, and lighting conditions, covering a variety of target postures; Use the annotation tool to mark the target in the screenshot.
6. The method according to claim 1, characterized in that: Enhanced processing of game screenshots includes: Adjusting the brightness and contrast of the game screenshots to adapt to different game environments; and / or Rotate, crop, and scale the game screenshots to improve model robustness; and / or Performing a color transformation on the game screenshots to prevent the model from overfitting to a specific game color style; and / or The game screenshots are mixed to enhance target diversity and improve generalization ability.
7. The method according to claim 1, characterized in that: The dynamically determining the priority of the shooting target based on the target distance, threat level and field of view angle includes: The priority of the shooting target is dynamically determined based on the following formula: Wherein, P is the priority of the shooting target, d, T, θ are the target distance, threat level, and field of view of the shooting target, respectively, w1, w2, and w3 are adjustable weight coefficients, and α, β, and γ are adjustment parameters used to adjust the impact of different factors on the priority.
8. The method according to claim 1, characterized in that: The method of predicting the position and speed of a high priority target based on Kalman filtering includes: Determining the current position and velocity of the high priority target based on Kalman filtering; The position and velocity of the high priority target in the next frame are predicted based on the following formula: v k+1 =υ k +aΔt Among them, x k and x k+1 are the current position and the position in the next frame of the high priority target, υ k and k+1 are the current speed and the speed in the next frame of the high priority target respectively; Δt is the time interval between two frames, and a is the acceleration.
9. The method according to claim 1, characterized in that: The mouse trajectory is a Bezier curve.
10. An automatic aiming device common to all game engines, characterized in that: include: A recognition module, used to obtain a game image, input the game image into a target recognition model to recognize a shooting target and return bounding box coordinates, wherein the target recognition model is generated by training a lightweight YOLO model with a data set consisting of enhanced game screenshots obtained under various conditions; A prediction module, for dynamically determining the priority of the shooting target based on the target distance, threat level and field of view angle, and predicting the position and speed of high priority targets based on Kalman filtering; The shooting module is used to generate a mouse trajectory that conforms to human behavior characteristics based on the position and speed of the high-priority target, and use hardware-level input to simulate the natural mouse movement trajectory to complete the shooting action at the target position.