Live-action model construction method and device taking entity target as minimum unit
A reality model and minimum unit technology, applied in the field of reality modeling, can solve the problems of high model construction cost and time-consuming, and achieve the effect of improving efficiency, low cost, and easy deployment
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0036] refer to figure 1 , the real scene model construction method of the present embodiment with the entity target as the smallest unit, which includes:
[0037] S101: Acquire an image of a target entity in the scene, and perform instance segmentation on the target entity in the scene.
[0038] In the specific implementation, the two-dimensional color image of the scene is obtained, and the instance segmentation algorithm based on deep learning is used to realize the segmentation of the target entity.
[0039] It is understandable that the commonly used deep learning models for instance segmentation include Mask R-CNN, MaskScoring R-CNN, TensorMask, SOLO, BlendMask, YOLACT, etc. Among them, YOLACT can achieve a segmentation speed of 33FPS, which can achieve real-time effects.
[0040] In this embodiment, the YOLACT deep learning model is used to perform instance segmentation on the target entity in the scene. In this way, the result of instance segmentation can obtain the...
Embodiment 2
[0066] This embodiment provides a device for constructing a reality model with a physical object as the smallest unit, which includes:
[0067] Instance segmentation module, which is used to obtain the image of the target entity in the scene, and carry out instance segmentation to the target entity in the scene;
[0068] A model calling module, which is used to identify the category of the target entity and call the model of the target entity from the entity target database;
[0069] A coordinate transformation module, which is used to extract the key points of the target entity model and convert the pixel coordinates of the key points into three-dimensional space coordinates;
[0070] A space occupancy calculation module, which is used to calculate the position and space occupancy of the target entity based on the three-dimensional space coordinates of the key points;
[0071] The model embedding module is used to embed the target entity model into the scene model based on t...
Embodiment 3
[0081] This embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps in the method for constructing a reality model with an entity object as the smallest unit as described in the first embodiment above are realized .
[0082] This embodiment aims at the scene where the physical target model is known, by calculating the position and space occupancy information of the target entity in the scene model, to realize the fast construction of the real scene model; directly using the model of the physical target, eliminating the need for 3D point cloud reconstruction Time consumption and computational complexity can improve the efficiency of real scene model construction; at the same time, only two-dimensional images of the scene need to be collected, no other equipment is required, and the cost is low and deployment is easy.
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More