Full-automatic travel photographing unmanned aerial vehicle system and method based on cloud AI

Abstract

The invention belongs to the technical field of unmanned aerial vehicles, and particularly relates to a cloud AI-based full-automatic travel photographing unmanned aerial vehicle system and method. The system comprises an unmanned aerial vehicle terminal which is carried by a user, does not comprise an AI chip and a cellular communication module, and is used for establishing a WiFi hotspot and performing photo and / or video shooting; the intelligent terminal APP is used for collecting voice or character instructions of a user, connecting the WiFi of the unmanned aerial vehicle and carrying out data transfer; and the cloud AI agent is used for analyzing the user instruction, calling a built-in photographer library to generate a photographing scheme, and realizing automatic picture retouching and video editing. According to the invention, the intelligent processing function is moved to the cloud, so that the hardware cost of the unmanned aerial vehicle is greatly reduced, and the functions of voice control, automatic shooting, intelligent picture retouching, video editing, a mirror-moving template, a picture style template, multi-person group photo, picture quality enhancement, one-key film formation, intelligent return flight, user preference learning and the like are realized at the same time; a common user can obtain high-quality travel shooting works without professional learning, and the method has the advantages of being easy to operate, low in cost, good in shooting effect and the like.

Description

[0001] This invention belongs to the field of drone technology, specifically relating to a drone system and method for intelligent shooting and intelligent image editing via cloud AI. Background Technology

[0002] Existing drones have the following problems:

[0003] 1. High price: Each drone needs to have a built-in AI chip and 4G / 5G module, resulting in high costs.

[0004] 2. Complex operation: It requires specialized learning of flight skills and photography knowledge, making it difficult for ordinary people to quickly master.

[0005] 3. The quality of the photos depends on the individual's skill level: Even if you can fly, if you don't understand professional photography techniques such as composition and camera movement, the quality of the photos and videos you take will be average.

[0006] 4. Not usable by a single person: When traveling alone, there is no one to help operate the drone to take photos or videos for you.

[0007] 5. Troublesome post-production: After shooting, you still need to edit the photos and videos yourself, which is a tedious process. Summary of the Invention

[0008] The technical problems this invention aims to solve are: existing drones are either expensive (due to built-in AI chips), complex to operate (requiring professional training), unusable by a single person (no one can help with filming), or require cumbersome post-processing (requiring manual image editing). This invention solves these problems simultaneously by moving AI to the cloud, allowing users to carry the drone themselves, supporting direct voice / text control, incorporating a built-in library of master photographers, and providing layered image editing optimization. This results in a travel photography drone that is "cheap, simple, professional, and usable by a single person."

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A fully automated travel photography drone system based on cloud-based AI, comprising:

[0011] The drone terminal, carried by the user, does not contain a dedicated chip for performing AI calculations or a cellular communication module for accessing the public network. The drone terminal is used to establish a local area network (LAN) Wi-Fi hotspot and perform flight photography tasks, including taking photos and / or videos. The drone terminal includes hardware such as a camera, microphone, audio playback module, Wi-Fi module, battery, and motors.

[0012] The smart terminal app, installed on the user's smart terminal device, is used to collect the user's voice and / or text commands, connect to the local area network wireless hotspot established by the drone, and relay data between the drone and the cloud-based AI agent. The app supports manual control, allowing users to control the drone in real time via screen operation.

[0013] The cloud-based AI agent, deployed on a remote server, receives user input commands forwarded by the smart terminal APP, calls the built-in photography master library to generate a shooting plan that includes flight trajectory, shooting parameters, and user guidance information, and converts the plan into executable commands for the drone, which are then returned to the drone via the smart terminal APP.

[0014] Furthermore, the cloud-based AI agent also includes:

[0015] The shooting plan generation module has a built-in photography master library, which includes a composition rule library, a camera movement trajectory library, a lighting technique library, and a posing guidance library. It is used to generate a shooting plan by calling the corresponding rules from the photography master library according to the voice command and / or environmental information.

[0016] The final video processing module automatically optimizes raw footage captured by the drone, generating optimized final videos. The raw footage includes photos and / or videos. This module allows users to submit modification requests for specific final videos via natural language input, such as "brighten photo 2," "shorten video 1 to 10 seconds," or "remove extra people from photo 1," and optimizes accordingly. The final video processing module employs a layered optimization strategy: for the first received modification request, non-geometric deformation optimization methods are prioritized, including lighting adjustments, color optimization, and contrast enhancement; when a second modification request is received, geometric deformation optimization methods are then used, including shape optimization, removal of extra people, and background replacement; and multi-round iteration is supported, allowing users to submit multiple modification requests consecutively, with the system responding sequentially.

[0017] The user preference learning module records the user's historical operation information, including manual adjustments, image editing commands, and final selections. In subsequent shooting scheme generation or final image processing, adjustments are made based on the historical operation information to achieve personalized adaptation.

[0018] The burst shooting optimization module is used to respond to the user's shooting command, control the drone to capture multiple images or a video, and select the best images in terms of composition, clarity, expression and other indicators to push to the user.

[0019] The connection methods are as follows: communication between the drone and the smart terminal APP is via the drone's self-built local area network (WiFi); communication between the smart terminal APP and the cloud-based AI agent is via the smart terminal device's wireless public network.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. Low cost: Drones do not require built-in AI chips and cellular modules, significantly reducing hardware costs.

[0022] 2. Simple to operate: Users can complete the shooting with just voice commands, without needing to learn professional skills.

[0023] 3. Excellent shooting results: The cloud-based AI has a built-in library of photography masters that can automatically plan the optimal composition and camera movement, suitable for both photos and videos.

[0024] 4. Intelligent Image Editing: Supports voice-activated image editing, video editing, layered optimization, and multiple iterations, resulting in natural and seamless image editing effects.

[0025] 5. Personalized learning: AI can record user preferences and learn more about users the more it is used.

[0026] 6. Wide range of applicable scenarios: It can be used in manual mode in areas with no signal, and automatically switches to full intelligent mode when there is a signal.

[0027] 7. Flexible interaction: Supports multiple modes such as automatic, manual, assisted correction, and template-based shooting.

[0028] 8. Comprehensive Functionality: Integrates extended functions such as group photos, image quality enhancement, one-click photo creation, intelligent return to home, and photo style templates to meet diverse needs.

[0029] Compared with existing centralized deployment and platform leasing drone solutions, this invention adopts a user-owned model, which does not rely on specific venues or operating platforms, and users can use it independently anytime and anywhere.

[0030] Compared to solutions that use local terminal AI processing, this invention places all intelligent computing in the cloud, further reducing the hardware cost of drones while improving AI processing capabilities and iteration speed.

[0031] Compared with traditional intelligent aerial photography systems, this invention enables ordinary users without professional backgrounds to consistently obtain high-quality shooting and post-processing effects through a dedicated library of master photographers and a layered optimization and retouching mechanism.

[0032] The photography master library and layered optimization retouching strategy proposed in this invention can effectively improve the professionalism of shooting and the effect of post-processing. The overall solution has outstanding practicality and novelty. Attached Figure Description

[0033] Figure 1 This is a system architecture diagram of the present invention.

[0034] Figure 2 This is an interactive flowchart of the present invention. Detailed Implementation

[0035] The invention will be further described below in conjunction with specific application scenarios.

[0036] Example 1: Fully Automatic Mode with Network Signal - Taking Photos

[0037] Scenario: A user is traveling alone in Huangshan and wants to take a photo in front of the Welcoming Pine.

[0038] Step 1: Place the drone on flat ground and turn on the power switch. After the drone starts up, it will automatically create a local area network WiFi hotspot named "Wingshot 1_XXXX".

[0039] Step 2: The user opens the "Yipaiyi" APP on their smart terminal and clicks to connect to the WiFi hotspot. After a successful connection, the APP interface displays "Drone connected".

[0040] Step 3: The user speaks into the microphone of the smart device: "Take a picture, based on the current environment, help me take a set of great photos." The app collects the voice command through the microphone.

[0041] Step 4: The app uploads the voice command to the cloud-based AI agent via a public wireless network. The cloud-based AI performs voice recognition and intent understanding, identifying that the user wants to take a "blockbuster" video, with the scene being the "Welcoming Pine".

[0042] Step 5: The cloud-based AI calls upon the built-in library of photography masters to generate a shooting plan: the drone takes off and moves 8 meters in front of the user at a height of 2.5 meters, locking the camera onto the subject; at the same time, it generates posing guidance voice: "Please take two steps to the left and look back at the camera."

[0043] Step 6: The cloud-based AI forwards flight commands and posture guidance voice prompts to the drone via the app. The drone takes off and performs its flight mission, simultaneously playing posture guidance voice prompts for the user.

[0044] Step 7: The drone automatically takes multiple photos, and after the photos are taken, the original materials are uploaded to the cloud AI via the APP.

[0045] Step 8: Cloud AI automatically retouches the photos, including color correction and composition optimization, and generates a finished image which is then pushed to the app. The app interface displays "Photo 1, Photo 2, Photo 3".

[0046] Step 9: The user views the final image and says into the microphone, "Brighten photo 2 a bit." The app captures the voice and uploads it to the cloud AI. The cloud AI optimizes the brightness of photo 2 and sends back a new version. Once the user is satisfied, they say, "Save this one."

[0047] [Example 2: Fully Automatic Mode with Network Signal - Video Recording]

[0048] Scenario: A user is at the beach and wants to film a video of themselves running.

[0049] Steps one through four are the same as above.

[0050] Step 5: The cloud-based AI calls upon the built-in library of photography masters to generate a shooting plan: The drone takes off and moves to the user's side and front, circling the user with the camera as the center, while generating posing guidance voice: "Please run forward, I will follow and shoot."

[0051] Steps 6 to 8: The drone performs filming, generates video footage, and cloud AI automatically edits, colors, and adds music to create a short video that is then pushed to the app.

[0052] Step 9: The user views the video and says into the microphone, "Cut the video down to 10 seconds." The cloud-based AI re-edits the video and sends back a new version.

[0053] [Example 3: Manual Mode Without Network Signal]

[0054] Scenario: A user goes on a trip to a remote mountain area, and their mobile phone has no public Wi-Fi signal.

[0055] Step 1: Place the drone on flat ground and turn on the power switch. The drone will automatically create a local area network (Wi-Fi) hotspot after starting up.

[0056] Step 2: The user opens the smart terminal app and connects to the WiFi hotspot. The app detects that it cannot connect to the cloud server and automatically prompts "No network currently available, switched to manual mode".

[0057] Step 3: Users can manually control the drone's flight using the virtual joystick on the app screen and control the taking of photos or videos using the shutter button on the screen.

[0058] Step 4: After shooting, the original footage is temporarily stored in the local storage of the smart device.

[0059] Step 5: The user returns to a location with a public Wi-Fi signal, opens the app, selects the "Upload to Cloud for Image Editing" function, and uploads the local materials to the cloud AI intelligent agent.

[0060] Step 6: The cloud-based AI agent automatically optimizes the footage, generates a finished video, and pushes it to the app, where users can view, save, or make further modifications.

[0061] [Example 4: Template-based shooting mode - camera movement template]

[0062] Scenario: A user is at the beach and wants to shoot a video with a "growing up onion" effect.

[0063] Step 1: The user opens the smart terminal APP and enters the "Camera Movement Templates" interface. The interface displays multiple template icons, including: "Up from the ground", "Circling Flight", "Gradual Pulling Up", "Spiral Ascent", "From Far to Near", etc. Each template corresponds to a set of preset flight paths and shooting parameters.

[0064] Step 2: The user clicks on the "Dry Land Onion Pulling" template. The app sends this template selection command to the cloud-based AI agent via the public wireless network.

[0065] Step 3: After receiving the instruction, the cloud-based AI agent calls the camera movement parameters corresponding to "Uprooting from Dry Land" from the photography master library: the flight trajectory is vertical ascent, the lens angle gradually rises from low to high, the ascent speed is 2 meters per second, and the shooting mode is video.

[0066] Step 4: The cloud-based AI converts the above parameters into executable commands for the drone, which are then forwarded to the drone via WiFi through the app.

[0067] Step 5: After receiving the command, the drone takes off automatically and executes the preset vertical ascent camera movement. At the same time, the camera rises smoothly from low to high to capture a complete picture of the user from the feet to the head and then to the background, ultimately presenting a "rising from the ground" visual effect.

[0068] Step 6: After shooting, the original footage is uploaded to the cloud AI via the APP.

[0069] Step 7: The cloud-based AI automatically optimizes the video based on the post-production style corresponding to the "Dry Land Pulling Onions" template (such as enhancing contrast and adding a slight stretching effect), and generates the final video to be pushed to the APP.

[0070] Step 8: Users can view the final video. If satisfied, they can save and share it; if not, they can make modification requests via voice, such as "slow down" or "lower the camera." The cloud AI will adjust the parameters according to the requests and regenerate the final video.

[0071] [Example 5: Template-based shooting mode - photo style template]

[0072] Scenario: After taking photos at the beach, the user wants a "retro" effect.

[0073] Step 1: Users open the smart terminal app and enter the "Photo Style" interface. The interface displays multiple style template icons, including: "Retro Film", "Japanese Fresh", "Black and White Blockbuster", "Cinematic", "Dreamy Soft Light", and "High-End Gray".

[0074] Step 2: The user clicks on the "Retro Film" template. The app sends this template selection command to the cloud-based AI agent via a public wireless network.

[0075] Step 3: After receiving the instruction, the cloud-based AI agent calls the corresponding retouching parameters for "Retro Film" in the photography master library: warm tone, add graininess, reduce saturation, enhance contrast, and add vignetting effect.

[0076] Step 4: The cloud-based AI applies the above parameters to the original photo to generate a new retro-style photo, which is then pushed to the app.

[0077] Step 5: Users can check the results. If satisfied, they can save and share; if not, they can make minor adjustments via voice, such as "lighter graininess" or "warmer tone." The cloud AI will adjust the parameters according to the requirements and regenerate.

[0078] The photo style template library of this invention can be continuously expanded according to user needs, including but not limited to: portrait (retro film, Japanese fresh style, black and white blockbuster, cinematic feel, dreamy soft light, sophisticated gray), landscape (landscape blockbuster, high saturation style, ink painting style, cyan and orange tones), and creative (cyberpunk, candy colors, monochrome emphasis). Each template corresponds to a set of preset retouching parameters, which users can call with one click or fine-tune by voice.

[0079] [Example 6: Extended Functionality - Group Photos, Image Quality Enhancement, One-Click Image Generation, Automatic Return]

[0080] Scenario: A family of three is traveling at the beach and wants to film a family photo video.

[0081] Sub-example 6.1: Group photo mode

[0082] Users open the app and click on the "Group Photo" mode. After the drone takes off, cloud-based AI identifies three faces in the camera feed. The AI ​​automatically calculates the optimal shooting position and generates instructions: the drone moves back to a distance of 8 meters from the group, adjusts its altitude to 2.5 meters, ensuring all three are centered in the frame, and uses the rule of thirds composition, with the people occupying 1 / 3 of the frame and the seascape occupying 2 / 3. The drone performs flight adjustments while simultaneously playing a posing voice prompt: "Please move closer, look at the camera, and get ready to shoot." After the photo is taken, the original footage is uploaded to the cloud.

[0083] Sub-example 6.2: Image quality enhancement

[0084] Cloud-based AI enhances the image quality of the original footage by processes including noise reduction, super-resolution, deblurring, and dynamic range extension.

[0085] Sub-example 6.3: One-click film production

[0086] Cloud-based AI automatically edits the enhanced footage: extracting the most exciting 15 seconds from 5 minutes of footage, automatically matching music, adding transition effects and text, and generating a short video with music to be pushed to the app.

[0087] Sub-example 6.4: Automatic Return and Priority Management

[0088] During filming, the drone monitors battery level and signal status in real time. The system has a preset return-to-home priority: user-initiated return-to-home > low battery return-to-home > signal loss return-to-home. When the battery level drops below 15% or the signal is lost for more than 10 seconds, the system will force a return-to-home.

[0089] Example 7: Live Streaming Mode

[0090] Scenario: The user is a travel blogger who is live-streaming in front of the Welcoming Pine in Huangshan and wants a drone to film him.

[0091] Step 1: The user opens the smart terminal app and clicks the "Live" button. The app sends a streaming request to the live streaming platform to obtain the streaming address.

[0092] Step 2: The user speaks into the microphone, "Start live streaming, follow me." The app collects the voice command and sends it to the cloud AI.

[0093] Step 3: Cloud AI analyzes the instructions and calls the photography master library to generate a live streaming shooting plan: the drone takes off and moves to 5 meters in front of the user at a height of 2 meters, and the lens locks onto the user.

[0094] Step 4: The drone's camera captures a real-time video stream, which is then transmitted to the app via WiFi. The app then pushes the stream to a live streaming platform.

[0095] Step 5: Users can control the live stream view in real time via voice commands, such as "zoom out" or "circle around." The cloud-based AI responds in real time and adjusts the drone's flight path.

[0096] Step 6: After the live stream ends, cloud AI automatically generates a live stream replay video and pushes it to the APP.

Claims

1. A fully automated travel photography drone system based on cloud-based AI, characterized in that, include: The drone terminal is carried by the user and does not contain a dedicated chip for performing AI calculations or a cellular communication module for accessing the public network; the drone terminal is used to establish a local area network wireless hotspot and perform flight photography tasks, including taking photos and / or videos. The smart terminal APP is installed on the user's smart terminal device and is used to collect the user's voice commands and / or text commands, connect to the local area network wireless hotspot established by the drone, and relay data between the drone and the cloud AI agent. The cloud-based AI agent, deployed on a remote server, receives user input commands forwarded by the smart terminal APP, calls the built-in photography master library to generate a shooting plan that includes flight trajectory, shooting parameters, and user guidance information, and converts the plan into executable commands for the drone, which are then returned to the drone via the smart terminal APP.

2. The system according to claim 1, characterized in that, The drone also includes a microphone for collecting ambient sound or as a backup input method for voice commands.

3. The system according to claim 1 or 2, characterized in that, The drone terminal also includes an audio playback module for playing user guidance information generated by the cloud-based AI agent.

4. The system according to claim 1, characterized in that, The smart terminal APP includes a manual control module, which is used to generate control commands in response to the user's manual operation on the APP interface and send them to the drone via the local area network wireless hotspot.

5. The system according to claim 1, characterized in that, The cloud-based AI agent includes a shooting plan generation module, which has a built-in photography master library. The photography master library includes a composition rule library, a camera movement trajectory library, a lighting technique library, and a posing guidance library. It is used to generate a shooting plan by calling the corresponding rules from the photography master library according to the voice command and / or environmental information.

6. The system according to claim 1, characterized in that, The cloud-based AI agent includes a video processing module, which automatically optimizes the original footage captured by the drone to generate optimized videos. The original footage includes photos and / or videos.

7. The system according to claim 6, characterized in that, The final image processing module allows users to submit modification requests for specified final images via natural language input, and optimizes the final images based on the modification requests.

8. The system according to claim 7, characterized in that, The image processing module adopts a layered optimization strategy: for the first received modification request, non-geometric deformation optimization methods are given priority, including lighting and shadow adjustment, color optimization, and contrast enhancement; when a second modification request is received from the user, geometric deformation optimization methods are then adopted, including lengthening legs, slimming faces, removing passersby, and replacing the background; and it supports multiple iterations, allowing users to submit multiple modification requests in succession, with the system responding to each one.

9. The system according to claim 1, characterized in that, The cloud-based AI agent includes a user preference learning module, which records the user's historical operation information and makes adjustments based on the historical operation information in subsequent shooting scheme generation or final image processing.

10. The system according to claim 1, characterized in that, The cloud-based AI agent includes a burst shooting optimization module, which responds to the user's shooting command, controls the drone to capture multiple images or a video, and selects the best footage to push to the user.

11. The system according to claim 1, characterized in that, The smart terminal APP also includes a template selection module, which is used to display multiple camera movement templates to the user. Each template corresponds to a set of preset flight trajectories and shooting parameters. In response to the user's selection command for the target template, the selection command is sent to the cloud AI agent, which then calls the corresponding camera movement parameters in the photography master knowledge base to generate flight control commands.

12. A fully automated travel photography method based on the system described in any one of claims 1-11, characterized in that, Includes the following steps: Establish a local area network wireless hotspot from the drone's end; The smart terminal APP connects to the local area network wireless hotspot; Users speak voice commands through the microphone of the smart terminal APP or enter text commands through the keyboard, and the smart terminal APP collects the user's input commands. The smart terminal APP forwards the user input commands to the cloud AI intelligent agent via a wireless public network; The cloud-based AI agent analyzes the user's input commands and calls upon the built-in library of photography masters to generate a shooting plan; The cloud-based AI agent converts the shooting plan into executable commands for the drone and forwards them to the drone via a smart terminal APP. The drone executes flight and photography tasks according to the executable instructions, and takes photos and / or videos; The drone uploads the raw footage to the cloud-based AI intelligent agent via a smart terminal app; The cloud-based AI agent optimizes the original materials, generates a finished product, and pushes it to a smart terminal APP. The finished product includes optimized photos and / or videos.

Images