Method for processing multimodal images, apparatus, device and storage medium

An apparatus for processing multimodal images, wherein the multimodal images are acquired using multiple types of vision sensors, and the multiple types of vision sensors are disposed in a first preset identity recognition scenario;

• the apparatus comprises:
• at least one processor;
• and
• a memory communicatively connected with the at least one processor;
• wherein,
• the memory stores thereon instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:
• if it is determined that a first vision sensor detects a biometric part of a target object, control each vision sensor to separately perform image acquisition for the biometric part in accordance with a preset acquisition strategy to obtain a target visual image of a corresponding type and acquisition time information of the target visual image;
• perform identity recognition for the target object according to a first target visual image to determine object identification information corresponding to the first target visual image;
• determine object identification information corresponding to a target visual image of other type other than the first target visual image according to the acquisition time information of each target visual image and the object identification information corresponding to the first target visual image;
• wherein the at least one processor is further enabled to:
• compare acquisition time information corresponding to the target visual image of other type with acquisition time information corresponding to the first target visual image;
• if an acquisition time corresponding to the target visual image of other type is within an acquisition time range corresponding to the first target visual image, determine the object identification information corresponding to the first target visual image to be the object identification information corresponding to the target visual image of other type.

A non-transitory computer-readable storage medium storing thereon computer instructions which are used to cause a computer to:

• if it is determined that a first vision sensor detects a biometric part of a target object, control each vision sensor to separately perform image acquisition for the biometric part in accordance with a preset acquisition strategy to obtain a target visual image of a corresponding type and acquisition time information of the target visual image;
• perform identity recognition for the target object according to a first target visual image to determine object identification information corresponding to the first target visual image;
• determine object identification information corresponding to a target visual image of other type other than the first target visual image according to the acquisition time information of each target visual image and the object identification information corresponding to the first target visual image;
• wherein the computer instructions are further used to cause the computer to:
• after determining the object identification information corresponding to the target visual image of other type other than the first target visual image according to the acquisition time information of each target visual image and the object identification information corresponding to the first target visual image,
• label the target visual image of other type with the object identification information to serve as a model training sample;
• train a preset recognition model using the model training sample to obtain a recognition model trained to convergence;
• wherein the multiple types of vision sensors are disposed in a second preset identity recognition scenario, and the computer instructions are further used to cause the computer to:
• determine acquisition environment information of the second preset identity recognition scenario;
• and
• if the acquisition environment information indicates an environment with sufficient visible light, use the first vision sensor to acquire the first target visual image of the target object, and perform identity recognition for the target object according to the first target visual image;
• if the acquisition environment information indicates an environment with insufficient visible light, use other vision sensor to acquire the target visual image corresponding to the target object, and perform identity recognition for the target object according to the corresponding target visual image and the recognition model trained to convergence.