Unmanned aerial vehicle video stream Internet ultra-low time delay transmission system and method thereof

Abstract

The invention discloses an unmanned aerial vehicle video stream Internet ultra-low delay transmission system and method. The system comprises an unmanned aerial vehicle, a cloud service and a user terminal. The unmanned aerial vehicle comprises a flight camera and an acquisition terminal, and the acquisition terminal is a double-acquisition terminal; the cloud service comprises a plurality of streaming media protocols and is used for packaging different streaming media; the flight camera comprises a video source acquisition unit, an attitude control unit, an Al identification unit, a video coding unit, an HDMI video output unit and an RTSP video output unit. The acquisition terminal comprises an RTSP stream pulling unit, a decoding unit, an encoding unit, an encoding control unit, an rtc stream pushing unit, a flight control data agent transfer unit, an HDMI to USB acquisition card unit, a virtual camera acquisition unit, an Al identification rendering unit and an Al data packaging unit. And the streaming media output by the flight camera is pulled through the HDMI-to-USB acquisition card unit or the RTSP stream pulling unit. According to the invention, the end-to-end time delay is reduced.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an ultra-low delay transmission system and method for an unmanned aerial vehicle video stream Internet. Background Art

[0002] The drone video streaming Internet ultra-low latency transmission system is a video low-latency transmission solution that uses the webrtc protocol to carry 5G terminals on drones. To meet the needs of drone video streaming network connection, a dedicated 4G / 5G Internet access module is usually deployed on the drone. After the video stream and flight control data are transmitted to the 5G base station through the 5G module, they are connected to the server. The cloud service provides streaming media access services and flight control services for downstream users. To solve the routing NAT restrictions, it is usually necessary to add rtsp pull stream + rtsp push stream work on the drone side.

[0003] The video stream and flight control data of drones need to be transmitted over long distances on the Internet or a private network, which will face many challenges. ① HDMI network transmission challenge: HDMI is not suitable for long-distance network transmission. If it is converted to network transmission, it needs to face the problems of HDMI data format conversion, encoding and video stream transmission transformation. ② Challenges of drone camera IP digital stream: The drone camera supplier will restrict the drone video encoding strategy, increase additional compatibility costs and delays. RTSP stream transmission uses UDP, which causes packet loss during transmission. ③ Challenges of routing NAT security restrictions: For network security considerations, most routers use port-restricted cone NAT, which makes it difficult for the external network to access internal network data. Users need to actively pull streams to them to obtain videos. When facing NAT networks, additional pull and push work is required, which will increase latency. ④ Challenges of end-side AI video recognition and video synchronization: It is more reasonable to place AI recognition on the push end. Many drone cameras, from a business strategy perspective, either do not provide AI recognition result output or only provide AI results directly rendered on the video screen, so their flexibility is greatly limited. Summary of the invention

[0004] The present invention provides an ultra-low delay transmission system and method for unmanned aerial vehicle video stream Internet, which are used to reduce the end-to-end delay.

[0005] In a first aspect, the present invention provides an ultra-low latency transmission system for a drone video stream over the Internet, comprising a drone, a cloud service, and a user terminal; the drone comprises a flying camera and a collection terminal, and the collection terminal is a dual collection terminal; the cloud service comprises a plurality of streaming media protocols for encapsulating different streaming media;

[0006] The flying camera includes a video source acquisition unit, a posture control unit, an AI recognition unit, a video encoding unit, an HDMI video output unit and an RTSP video output unit;

[0007] The acquisition terminal includes an RTSP streaming unit, a decoding unit, an encoding unit, an encoding control unit, an RTC streaming unit, a flight control data proxy transfer unit, an HDMI to USB acquisition card unit, a virtual camera acquisition unit, an Al recognition rendering unit and an Al data encapsulation unit;

[0008] The streaming media output by the flying camera is pulled through the HDMI to USB acquisition card unit or the RTSP streaming unit.

[0009] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the AI ​​recognition rendering unit uses the optimized target detection model yolov8n and the target tracking deepsort model to perform AI recognition processing.

[0010] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the optimized target detection model yolov8n is a model obtained by integrating the self-correcting illumination network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network.

[0011] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, AI recognition processing is performed based on the optimized target detection model yolov8n and the target tracking deepsort model as follows:

[0012] Performing target detection on the video frames in the streaming media by using the optimized target detection model yolov8n to obtain the position of the target in the streaming media;

[0013] The target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked by the target tracking deepsort model to obtain the current frame ID;

[0014] The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

[0015] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the self-correcting lighting network SCINet includes a cascade layer, a weight sharing layer and a self-correcting layer.

[0016] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the cascade layer is used to perform illumination estimation on the image input at each stage through cascade illumination learning; the weight sharing layer is used to share model parameters; and the self-correction layer is used to correct the output of the current stage.

[0017] In a second aspect, the present invention further provides a method for transmitting drone video streams over the Internet with ultra-low latency, comprising:

[0018] Pull the streaming media output by the flying camera through the HDMI to USB acquisition card unit or the RTSP streaming unit;

[0019] The optimized target detection model yolov8n is used to perform target detection on the video frames in the streaming media to obtain the position of the target in the streaming media; the optimized target detection model yolov8n is obtained by integrating the self-correction illumination network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network;

[0020] The target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked through the target tracking deepsort model to obtain the current frame ID;

[0021] The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

[0022] In a third aspect, the present invention also provides an electronic device, comprising: a memory for storing a computer software program; and a processor for reading and executing the computer software program, thereby realizing the above-mentioned method for ultra-low-latency transmission of drone video streams over the Internet.

[0023] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium, wherein a computer software program is stored in the storage medium. When the computer software program is executed by a processor, the above-mentioned method for ultra-low-latency transmission of drone video streams over the Internet is implemented.

[0024] In a fifth aspect, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned method for ultra-low-latency transmission of drone video streams over the Internet.

[0025] The UAV video stream Internet ultra-low delay transmission system provided by the present invention comprises a UAV, a cloud service and a user terminal; the UAV comprises a flying camera and a collection terminal, and the collection terminal is a dual collection terminal; the cloud service comprises a plurality of streaming media protocols for encapsulating different streaming media; the flying camera comprises a video source collection unit, a posture control unit, an Al recognition unit, a video encoding unit, an HDMI video output unit and an RTSP video output unit; the collection terminal comprises an RTSP stream pulling unit, a decoding unit, an encoding unit, an encoding control unit, an RTC stream pushing unit, a flight control data proxy transfer unit, an HDMI to USB collection card unit, a virtual camera collection unit, an Al recognition rendering unit and an Al data encapsulation unit; the streaming media output by the flying camera is pulled through the HDMI to USB collection card unit or the RTSP stream pulling unit. Therefore, the embodiment of the present invention supports both HDMI input and RTSP streaming at the video source. Therefore, the HDMI output data can be converted into virtual camera data input, and its end-to-end delay is expected to be controlled within 120ms. Compared with the rtsp stream end-to-end delay starting at 200ms-600ms, there is a significant improvement. The rtsp streaming solution and HDMI are both upgraded to RTC protocol streaming. The RTC protocol is combined with the encoder, which can dynamically adjust the video bit rate and rtp packet size based on network delay, so as to obtain faster end-to-end delay. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a structural schematic diagram of the Internet ultra-low delay transmission system for drone video streams provided by the present invention;

[0027] Figure 2 It is the overall framework diagram of SCINet provided by the present invention;

[0028] Figure 3 It is a flow chart of the method for transmitting UAV video streams over the Internet with ultra-low latency provided by the present invention;

[0029] Figure 4 A schematic diagram of an electronic device according to an embodiment of the present invention;

[0030] Figure 5 A schematic diagram of an embodiment of a computer-readable storage medium provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0032] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0033] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0034] Optional, see Figure 1 , Figure 1 It is a structural schematic diagram of the UAV video stream Internet ultra-low latency transmission system provided by the present invention. Specifically, the UAV video stream Internet ultra-low latency transmission system includes a UAV, a cloud service and a user terminal; the UAV includes a flying camera and a collection terminal, and the collection terminal is a dual collection terminal; the cloud service includes a variety of streaming media protocols for encapsulating different streaming media.

[0035] Optionally, the flight camera includes a video source acquisition unit, a posture control unit, an Al recognition unit, a video encoding unit, an HDMI video output unit and an RTSP video output unit.

[0036] Optionally, the acquisition terminal includes an RTSP streaming unit, a decoding unit, an encoding unit, an encoding control unit, an RTC streaming unit, a flight control data proxy transfer unit, an HDMI to USB acquisition card unit, a virtual camera acquisition unit, an Al recognition rendering unit and an Al data encapsulation unit.

[0037] Optionally, use an HDMI to USB capture card unit or RTSP streaming unit to pull the streaming media output by the flight camera.

[0038] The UAV video stream Internet ultra-low delay transmission system of the embodiment of the present invention includes a UAV, a cloud service and a user terminal; the UAV includes a flying camera and a collection terminal, and the collection terminal is a dual collection terminal; the cloud service includes a plurality of streaming media protocols for encapsulating different streaming media; the flying camera includes a video source collection unit, a posture control unit, an Al recognition unit, a video encoding unit, an HDMI video output unit and an RTSP video output unit; the collection terminal includes an RTSP streaming unit, a decoding unit, an encoding unit, an encoding control unit, an RTC streaming unit, a flight control data proxy transfer unit, an HDMI to USB collection card unit, a virtual camera collection unit, an Al recognition rendering unit and an Al data encapsulation unit; the streaming media output by the flying camera is pulled through the HDMI to USB collection card unit or the RTSP streaming unit. Therefore, the embodiment of the present invention supports both HDMI input and RTSP streaming at the video source. Therefore, the HDMI output data can be converted into virtual camera data input, and its end-to-end delay is expected to be controlled within 120ms. Compared with the rtsp stream end-to-end delay starting at 200ms-600ms, there is a significant improvement. The rtsp streaming solution and HDMI are both upgraded to RTC protocol streaming. The RTC protocol is combined with the encoder, which can dynamically adjust the video bit rate and rtp packet size based on network delay, so as to obtain faster end-to-end delay.

[0039] For AI video stream recognition and tracking, according to the scene requirements, the improved lightweight target detection module yolov8n (SCINet) and the target tracking deepsort algorithm are used for AI recognition processing, taking into account both reasoning speed and accuracy. Therefore, the AI ​​recognition rendering unit in the embodiment of the present invention uses the optimized target detection model yolov8n and the target tracking deepsort model for AI recognition processing. Figure 2 , Figure 2 It is the overall framework diagram of SCINet provided by the present invention.

[0040] The core idea of ​​the YOLO series of target detection algorithms is to transform target detection into a regression problem, using the entire image as the input of the network, and obtaining the position of the bounding box and its category through the neural network. DeepSORT is a computer vision tracking algorithm that improves the accuracy and robustness of target tracking by combining deep learning with traditional tracking techniques (Kalman filtering and Hungarian algorithm). It is used to track objects while assigning an ID to each detected object to achieve real-time tracking effects.

[0041] In view of the difficulty of identifying and tracking drones in low light at night, the optimized target detection model yolov8n in the embodiment of the present invention is a model obtained by integrating the self-correction illumination network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network. Therefore, it can be understood that SCINet (self-correction illumination network) is integrated into the backbone network part of the benchmark yolov8n network to create a unique yolov8n (SCINet) + deepsort strategy. Therefore, by adopting this strategy, in the actual AI tracking scenario, the problem of difficulty in identifying and tracking videos under low light is solved, and the illumination part is optimized to improve image quality to complete night target tracking, showing its application value. In addition, SCINet improves image quality by means of cascaded illumination learning and weight sharing without increasing model complexity, so the improved method has a considerable speed compared with the benchmark yolov8n. Combining yolov8n (SCINet) + deepsort with drone scenes This method is mainly aimed at the field of drone tracking using a camera from a high altitude in low-light conditions such as dusk or night. It is suitable for using drones in open areas such as roads, plateaus, and plains to hunt suspects, track suspicious vehicles, and monitor and track traffic. In the above scenarios, drones may encounter different weather conditions (such as haze, cloudy, sunset, etc.) when performing tasks. The SCINet module uses image enhancement technology to help improve the target recognition performance under these conditions and ensure that the drone's mission is not limited by environmental factors (such as losing the tracking target, misidentifying the target, and other adverse effects). Due to the training of the DisDrone dataset, the targets for identification and tracking include 9 categories: pedestrians, bicycles, cars, vans, trucks, tricycles, awning tricycles, buses, and motorcycles. The identification type fully meets the tracking needs of scenes such as hunting suspects, tracking suspicious vehicles, and traffic monitoring and tracking. In general, after integrating SCINet into YOLOv8n, the target recognition ability of drones in low-light and complex environments can be significantly improved, providing more reliable support for various application scenarios. Deepsort can more accurately assign IDs to targets and achieve tracking under the premise of reliable target detection.

[0042] Therefore, the process of AI recognition processing based on the optimized target detection model yolov8n and the target tracking deepsort model is as follows: the optimized target detection model yolov8n is used to detect the target of the video frames in the streaming media to obtain the position of the target in the streaming media; the target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked through the target tracking deepsort model to obtain the current frame ID; the target position detected in each frame is updated, and the tracking frame is continuously associated with the target position to track the dynamic position of the target.

[0043] Therefore, through the specific implementation method of yolov8n (SCINet), the cascade illumination learning, weight sharing and self-correction modules in SCINet can improve the enhancement effect of low-light images and realize target detection of yolov8n (SCINet) in low light.

[0044] 1. Basic principles: This deep learning network uses a multi-level feature extraction method to capture image information of different scales by establishing a self-correcting framework, and automatically adjusts the lighting conditions of the input image. The features in the image are used to form parameter indicators to analyze the current lighting level and perform self-correction to enhance the visibility of the image. The contrast and brightness of the image are then optimized to make the dark details clearer, ensuring the accuracy of subsequent target detection. The yolov8n (SCINet) network is trained using the dedicated drone aerial photography dataset VisDrone2021 to obtain a yolov8n (SCINet) model that combines speed and accuracy. This dataset contains a large number of real-life pictures with different illumination levels, which fully demonstrates the low-light detection performance of the model.

[0045] 2. The self-correction illumination network SCINet in the embodiment of the present invention includes a cascade layer, a weight sharing layer and a self-correction layer, wherein the cascade layer is used to estimate the illumination of the image input at each stage through cascade illumination learning; the weight sharing layer is used to share model parameters; and the self-correction layer is used to correct the output of the current stage. Therefore, it can be understood that 1. Cascade process: In cascade illumination learning, the model consists of multiple stages, each of which estimates the illumination of the input image. This multi-stage processing method helps to gradually improve the brightness of the image, and each stage further improves the image quality based on the previous stage. 2. Weight sharing: In these stages, the parameters (weights) of the model are shared. This means that although each stage performs similar tasks, they use the same model parameters to perform these tasks. Weight sharing can reduce the overall number of parameters of the model, thereby reducing the complexity of the model and the risk of overfitting. 3. Self-correction module: After each stage, a self-correction module is used to correct the output of the current stage. This correction ensures that as the cascade process proceeds, the output of each stage gradually stabilizes, and the quality of the final output will not be degraded due to too many processing steps.

[0046] See also Figure 3 , Figure 3 A schematic diagram of a flow chart of a method for transmitting a drone video stream over the Internet with ultra-low latency provided by an embodiment of the present invention. The method for transmitting a drone video stream over the Internet with ultra-low latency includes:

[0047] Step 10: Pull the streaming media output by the flying camera through the HDMI to USB capture card unit or RTSP streaming unit;

[0048] Step 20, performing target detection on the video frames in the streaming media through the optimized target detection model yolov8n to obtain the position of the target in the streaming media;

[0049] Step 30, match the target position detected in the current frame with the processed tracking frame of the previous frame, and track the target through the target tracking deepsort model to obtain the current frame ID;

[0050] Step 40, updating the target position detected in each frame, and continuously associating the tracking frame with the target position to track the dynamic position of the target.

[0051] The human-machine video stream Internet ultra-low latency transmission method in the embodiment of the present invention is similar to the process of the human-machine video stream Internet ultra-low latency transmission system described above, and will not be repeated here. Among them, the optimized target detection model yolov8n is obtained by integrating the self-correction lighting network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network.

[0052] The ultra-low delay Internet transmission method for drone video streams provided by the present invention supports both HDMI input and RTSP stream at the video source. Therefore, HDMI output data can be converted into virtual camera data input, and its end-to-end delay is expected to be controlled within 120ms. Compared with the rtsp stream end-to-end delay starting at 200ms-600ms, it has a relatively huge improvement. The rtsp stream solution and HDMI are both upgraded to RTC protocol streaming. The RTC protocol is combined with an encoder, and the video bit rate and rtp packet size can be dynamically adjusted based on network delay, so as to obtain faster end-to-end delay.

[0053] See also Figure 4 , Figure 4 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention provides an electronic device 400, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, the following steps are implemented:

[0054] Performing target detection on the video frames in the streaming media by using the optimized target detection model yolov8n to obtain the position of the target in the streaming media;

[0055] The target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked by the target tracking deepsort model to obtain the current frame ID;

[0056] The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

[0057] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the self-correcting lighting network SCINet includes a cascade layer, a weight sharing layer and a self-correcting layer.

[0058] See also Figure 5 , Figure 5 Schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 5 As shown, this embodiment provides a computer-readable storage medium 500, on which a computer program 511 is stored. When the computer program 511 is executed by a processor, the following steps are implemented:

[0059] Performing target detection on the video frames in the streaming media by using the optimized target detection model yolov8n to obtain the position of the target in the streaming media;

[0060] The target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked by the target tracking deepsort model to obtain the current frame ID;

[0061] The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

[0062] According to the UAV video stream Internet ultra-low latency transmission system provided by the present invention, the self-correcting lighting network SCINet includes a cascade layer, a weight sharing layer and a self-correcting layer.

[0063] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0064] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0066] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0068] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0069] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An ultra-low latency transmission system for drone video streams over the Internet, characterized in that: It includes a drone, a cloud service and a user terminal; the drone includes a flying camera and a collection terminal, and the collection terminal is a dual collection terminal; the cloud service includes a variety of streaming media protocols for packaging different streaming media; The flying camera includes a video source acquisition unit, a posture control unit, an AI recognition unit, a video encoding unit, an HDMI video output unit and an RTSP video output unit; The acquisition terminal includes an RTSP streaming unit, a decoding unit, an encoding unit, an encoding control unit, an RTC streaming unit, a flight control data proxy transfer unit, an HDMI to USB acquisition card unit, a virtual camera acquisition unit, an Al recognition rendering unit and an Al data encapsulation unit; The streaming media output by the flying camera is pulled through the HDMI to USB acquisition card unit or the RTSP streaming unit.

2. The UAV video stream Internet ultra-low delay transmission system according to claim 1 is characterized in that: The AI ​​recognition rendering unit uses the optimized target detection model yolov8n and the target tracking deepsort model to perform AI recognition processing.

3. The UAV video stream Internet ultra-low delay transmission system according to claim 2 is characterized in that: The optimized target detection model yolov8n is a model obtained by integrating the self-correcting illumination network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network.

4. The UAV video stream Internet ultra-low delay transmission system according to claim 3 is characterized in that: The process of AI recognition processing based on the optimized target detection model yolov8n and target tracking deepsort model is as follows: Performing target detection on the video frames in the streaming media by using the optimized target detection model yolov8n to obtain the position of the target in the streaming media; The target position detected in the current frame is matched with the processed tracking frame of the previous frame, and the target is tracked by the target tracking deepsort model to obtain the current frame ID; The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

5. The UAV video stream Internet ultra-low latency transmission system according to claim 3 or 4, characterized in that: The self-correction illumination network SCINet includes a cascade layer, a weight sharing layer and a self-correction layer.

6. The UAV video stream Internet ultra-low delay transmission system according to claim 5 is characterized in that: The cascade layer is used to estimate the illumination of the image input in each stage through cascade illumination learning; the weight sharing layer is used to share model parameters; and the self-correction layer is used to correct the output of the current stage.

7. A method for transmitting drone video streams over the Internet with ultra-low latency, characterized in that: include: Pull the streaming media output by the flying camera through the HDMI to USB acquisition card unit or the RTSP streaming unit; Performing target detection on the video frames in the streaming media by using the optimized target detection model yolov8n to obtain the position of the target in the streaming media; The optimized target detection model yolov8n is obtained by integrating the self-correction illumination network SCINet into the backbone network part of the benchmark lightweight target detection model yolov8n network; Match the target position detected in the current frame with the processed tracking frame of the previous frame, and track the target through the target tracking deepsort model to obtain the current frame ID; The detected target position is updated in each frame, and the tracking box is continuously associated with the target position to track the dynamic position of the target.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for ultra-low latency transmission of drone video streams over the Internet as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the method for ultra-low-latency transmission of drone video streams over the Internet as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by the processor, the method for ultra-low-latency transmission of drone video streams over the Internet as described in any one of claims 1 to 6 is implemented.

Images