Stmicroelectronics S.r.l.

Stmicroelectronics (rousset) Sas

Stmicroelectronics (grenoble 2) Sas

Mirko Falchetto

Danilo Pietro Pau

Laurent Folliot

Emanuele Plebani

Pierre Demaj

Massimiliano Barone

Paolo Pasteris

According to an aspect, a method is proposed for defining placements, in a volatile memory, of temporary scratch buffers used during an execution of an artificial neural network, the method comprising: determining an execution order of layers of the neural network, defining placements, in a heap memory zone of the volatile memory, of intermediate result buffers generated by each layer, according to the execution order of the layers, determining at least one free area of the heap memory zone over the execution of the layers, defining placements of temporary scratch buffers in the at least one free area of the heap memory zone according to the execution order of the layers.

Method for memory allocation during execution of a neural network

According to one aspect, a method for determining, for a memory allocation, placements in a memory area of data blocks generated by a neural network, comprises a development of an initial sequence of placements of blocks, each placement being selected from several possible placements, the initial sequence being defined as a candidate sequence, a development of at least one modified sequence of placements from a replacement of a given placement of the initial sequence by a memorized unselected placement, and, if the planned size of the memory area obtained by this modified sequence is less than that of the memory area of the candidate sequence, then this modified sequence becomes the candidate sequence, the placements of the blocks for the allocation being those of the placement sequence defined as a candidate sequence once each modified sequence has been developed.

Device and method for allocating intermediate data from an artificial neural network

Methods, microprocessors, and systems are provided for implementing an artificial neural network. Data buffers in virtual memory are coupled to respective processing layers in the artificial neural network. An ordered visiting sequence of layers of the artificial neural network is obtained. A virtual memory allocation schedule is produced as a function of the ordered visiting sequence of layers of the artificial neural network, the schedule including a set of instructions for memory allocation and deallocation operations applicable to the data buffers. A physical memory configuration dataset is computed as a function of the virtual memory allocation schedule for the artificial neural network, the dataset including sizes and addresses of physical memory locations for the artificial neural network.

Automatic memory management method, corresponding micro-controller unit and computer program product

Image-processing apparatus and methods to adaptively vary an interest point threshold value and control a number of interest points identified in an image frame are described. Sub-regions of an image frame may be processed in a sequence, and an interest point threshold value calculated for each sub-region. The calculated value of the interest point threshold may depend upon pre-selected values and values determined from the processing of one or more prior sub-regions. By using adaptive thresholding, a number of interest points detected for each frame in a sequence of image frames may remain substantially constant, even though objects within the frames may vary appreciably.

Adaptive interest rate control for visual search

Image-processing apparatus and methods to adaptively control a size and/or location of a visual search window used for feature matching in a machine-vision system are described. A search window controller may receive motion vector data and image recognition rate data, and compute a search window size and/or search window location based on the received data. The computed search window size may be a portion of an image frame. The motion vector data and image recognition rate data may be computed from one or more images in a video image sequence. By adaptively controlling search window size and location, an appreciable reduction in data processing burden for feature matching may be achieved.

Adaptive search window control for visual search

A graphic system having a central processing unit; a system memory coupled to the central processing unit; a display unit provided with a corresponding screen; a graphic module coupled to and controlled by the central processing unit to render an image on the screen of the display unit, the graphic module including a fragment graphic module having a depth test buffer for storing a current depth value; a depth test stage coupled to the depth test buffer for comparing the current depth value with a depth coordinate associated with an incoming fragment and defining a resulting fragment; a test stage for testing the resulting fragment and defining a retained fragment; a buffer writing stage operatively associated with the test stage for receiving the retained fragment, the buffer writing stage coupled to the depth test buffer for updating the current depth value with a depth value of the retained fragment.

Graphic system comprising a fragment graphic module and relative rendering method

A method for rendering a three dimensional scene on a displaying screen comprises: generating for a tile of a current scene a hierarchical z-buffer which comprises a plurality of levels organized according to depth values; calculating a minimum depth value d of a submitted primitive; calculating an intersection area associated with said primitive with respect to said tile; providing a multiplicity of aligned regions each associated with a level of the hierarchical z-buffer so that the exact area calculated is suitable to be covered, at least entirely, by the union of such aligned regions; comparing the minimum depth value d of the submitted primitive with corresponding maximum depth values v, v, . . . , vN each read from the levels of the hierarchical z-buffer; discarding said primitive whether the minimum depth value d is bigger than all maximum depth values v, v, . . . , vN.

Graphic rendering method and system comprising a graphic module

The disclosure relates to a graphics module for rendering a bidimensional scene on a display screen comprising a graphics pipeline of the sort-middle type, said graphics pipeline comprising: a first processing module configured to clip a span-type input primitive received from a rasterizer module into sub-span type primitives to be associated to respective macro-blocks corresponding to portions of the screen, and to store said sub-span type primitives in a scene buffer; a second processing module configured to reconstruct the span-type input primitive starting from said sub-span type primitives, the second processing module being further intended to implement a culling operation of sub-span type primitives of the occluded type.

Rendering module for bidimensional graphics

A graphic module wherein, given a curve P(x,y)=P(x(t), y(t)) between two points P(x,y)=P(x(t), y(t)) and P(x,y)=P(x(t), y(t), in the screen coordinates, a calculating circuit computes a curve mid-point P(x,y)=P(x(t), y(t)), where t=(t+t)/2 on the curve, computes a segment mid-point P(x,y)=(P(x,y)+P(x,y))/2 on segment, computes a distance function d between the curve mid-point and the segment mid-point, and, given two thresholds THRand THR, with THR&#x3c;=THR, if d&#x3c;THR, it generates line segment, and if THR&#x3c;=d&#x3c;THR, it generates two line segmentsandif d&#x3e;=THR, it repeats the previous steps for the curve portions from Pto Pand from Pto P.

System and method for adaptive tessellation of a curve

A graphic rendering method includes: providing data primitive representing primitives of a scene; defining a plurality of three-dimensional cells of a scene view frustum; ordering the cells according to an order based on cell depths from a reference plane; associating each primitive to a cell; and processing data primitives according to the cell order to renderize the scene.

Growing community of inventors

Milzano, Italy

Mirko Falchetto