An image processing system with scalable models is provided. The image processing system comprises computing devices having a graphic analysis environment that includes instructions to execute an analysis process on a first image having a native resolution. The analysis process causes the one or more computing devices to perform operations includes: resampling the first image to generate a second image, wherein the second image has a resampled resolution greater than the native resolution in pixel number; detecting a plurality of first patches and a plurality of second patches in the first image and the second image, respectively, wherein the first patches and the second patches are detected by different detection models of a first scalable model collection according to sizes of the first image and the second image; and aggregating the first patches and the second patches. A method for processing an image with scalable models is also provided.

Methods, systems, and apparatus, including medium-encoded computer program products, for a secure edge platform that uses image classification machine learning models. An edge platform can include at least one camera and can identify image classification models that generate classification output data from image data generated by the cameras. The edge platform can receive image data generated by the camera, and provide the image data to the models. In response to providing the image data classification models, the edge platform can receive classification output data. In response to receiving the classification output data from the image classification models, the edge platform can generate augmentation data that is associated with the image data, then transmit detection data to a central server platform. The detection data can include (i) the classification output data and (ii) the augmentation data associated with the image data. Data can be made recordable, reportable, searchable, and alarmable.

Methods and systems are described herein for hosting and arbitrating algorithms for the generation of structured frames of data from one or more sources of unstructured input frames. A plurality of frames may be received from a recording device and a plurality of object types to be recognized in the plurality of frames may be determined. A determination may be made of multiple machine learning models for recognizing the object types. The frames may be sequentially input into the machine learning models to obtain a plurality of sets of objects from the plurality of machine learning models and object indicators may be received from those machine learning models. A set of composite frames with the plurality of indicators corresponding to the plurality of objects may be generated, and an output stream may be generated including the set of composite frames to be played back in chronological order.

The present disclosure relates to an information processing apparatus, an information processing method, an information processing system, and a program capable of appropriately evaluating an object recognition filter by simpler processing. A generation unit that generates teacher data of a preprocessing filter provided in a preceding stage of the object recognition filter is generated by a cyclic generative adversarial network (Cyclic GAN) that is unsupervised learning. The teacher data generated by the generated generation unit is applied to the object recognition filter, an evaluation image is generated from a difference between object recognition result images, and an evaluation filter that generates an evaluation image from the evaluation image and the teacher data is generated. The evaluation filter is applied to an input image to generate an evaluation image, and the object recognition filter is evaluated by the generated evaluation image. The present disclosure can be applied to an object recognition device.

A novel and useful system and method of improved power performance and lowered memory requirements for an artificial neural network based on packing memory utilizing several structured sparsity mechanisms. The invention applies to neural network (NN) processing engines adapted to implement mechanisms to search for structured sparsity in weights and activations, resulting in a considerably reduced memory usage. The sparsity guided training mechanism synthesizes and generates structured sparsity weights. A compiler mechanism within a software development kit (SDK), manipulates structured weight domain sparsity to generate a sparse set of static weights for the NN. The structured sparsity static weights are loaded into the NN after compilation and utilized by both the structured weight domain sparsity mechanism and the structured activation domain sparsity mechanism. The application of structured sparsity lowers the span of search options and creates a relatively loose coupling between the data and control planes.

The purpose of the present invention is to provide a technology for updating a recognition model so that even if there were errors in recognition of an unknown scene or the like, the scene can be recognized quickly. The present invention is provided with a data analysis unit 11 that, on the basis of data from an outside recognition unit 32 provided to a vehicle, acquires from among previously stored recognition models a model approximate to a recognition model recognized by the outside recognition unit 32, and that reproduces the acquired model in the form of computer graphics images. The data analysis unit 11 is provided with: a difference extraction unit 114 that compares the reproduced computer graphics images and data from the outside recognition unit 32 and extracts a difference therebetween; an object recognition unit 116 that recognizes an object relating to the difference extracted by the difference extraction unit 114; and a scene reconfiguration unit 117 that creates computer graphics images having the object recognized by the recognition unit 116 reflected therein.

A method, an apparatus, a system, and an electronic device for selecting an intelligent analysis algorithm. The method includes: acquiring image data of a monitoring scene (S101); analyzing the image data to obtain scene contents contained in the image data (S102); determining an intelligent analysis algorithm corresponding to each of the scene contents (S103); and selecting a target intelligent analysis algorithm(s) from intelligent analysis algorithms corresponding to the scene contents according to a load capacity of a compute node used for loading the intelligent analysis algorithms, wherein a total algorithm load of the target intelligent analysis algorithm(s) is not greater than the load capacity of the compute node (S104). The method for selecting an intelligent analysis algorithm realizes an automatic selection of the intelligent analysis algorithm, which can reduce the manual workload, improve the selection efficiency of the intelligent analysis algorithm, reduce overload of the compute node, reduce abnormal analysis results caused by the overload of the compute node, and reduce an improper selection of the intelligent analysis algorithm due to the low degree of professionalism of the construction personnel, which affects the analysis effect.

A training method for an image fusion processing model is provided. The method includes: obtaining an image set, and compressing the image set, updating a parameter of an encoder of a single image processing model and a parameter of a decoder of the single image processing model according to a single to-be-replaced face in the original image set, and updating parameters of an encoder and a decoder that are of the image fusion processing model according to different to-be-replaced faces and different target faces that are in the original image set while the parameters of the encoder and the decoder that are of the single image processing model remain unchanged. An image processing method and apparatus for an image fusion processing model, an electronic device, and a storage medium are further provided.

A media source is configured to provide a media stream that includes an audio-video recording. A first algorithm detects a presence of synthetic media in a video based on a first feature type. A second algorithm detects the presence of synthetic media based on a second feature type. A device determines that the first feature type is presented in a selected portion of the image frames but the second feature type is not. In response, the first algorithm is executed using the selected portion of the image frames as an input. Based on the executed first algorithm, a probability is determined that the media stream provided by the media source includes synthetic media.

In a facial skin detection method performed by a terminal device having a front-facing camera, a first image including a face is obtained and feature information of pores in a target area of the face is extracted. The pores in the target area are classified into pore categories based on the feature information of the pores in the target area. Feature information of pores of each pore category is input into at least one pore detection model to obtain pore detection data of each pore category and a skin detection result of the face is determined based on pore detection data of the pore categories.