Methods, systems and apparatus to improve image classification with boundary-bitmaps

Abstract

Methods, systems, apparatus and articles of manufacture are disclosed herein to improve image classification with boundary-bitmaps. An example disclosed apparatus includes a silhouette engine to identify a foreground silhouette within the image, generate a bounding box based on borders of the foreground silhouette, and generate an encoded silhouette matrix which identifies cells of a foreground and cells of a background, a convolution cell selector to convolve the encoded silhouette matrix to generate a convoluted bitmap matrix, and a filter cell selector to improve image classification efficiency by identifying eligible blocks of the convoluted bitmap matrix by retaining first respective cells of the convoluted bitmap matrix that satisfy a cell retention threshold, and removing second respective cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold.

1 - 35 . (canceled) 
     
     
         36 . An apparatus to improve image classification efficiency, the apparatus comprising:
 a silhouette engine to generate an encoded silhouette matrix that includes a bitmap area defined by transitions between foreground image cells and background image cells;   a convolution cell selector to reduce a quantity of the foreground image cells within the bitmap area to be processed by convolving the encoded silhouette matrix to generate a convoluted bitmap matrix; and   a filter cell selector to:
 select first cells of the convoluted bitmap matrix that satisfy a cell retention threshold value; 
 select second cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold value; 
 generate a contributed blocks matrix by encoding the first cells with respective retention indicators; and 
 encode the second cells with respective discard indicators, the second cells to be removed from processing by an image classifier. 
   
     
     
         37 . The apparatus as defined in  claim 36 , further including a boundary encapsulator to generate an eligible block matrix by identifying an outer perimeter based on the respective retention indicators. 
     
     
         38 . The apparatus as defined in  claim 37 , wherein the boundary encapsulator is to replace cells internal to the outer perimeter with respective discard indicators, the cells internal to the outer perimeter indicative of a portion of the foreground cells. 
     
     
         39 . The apparatus as defined in  claim 38 , further including a classification interface circuit to mask a detection window with the eligible block matrix. 
     
     
         40 . The apparatus as defined in  claim 36 , further including a window engine to:
 convolve a ones matrix kernel to respective cells of the encoded silhouette matrix; and   calculate a sum total of the foreground cells within the ones matrix kernel corresponding to the encoded silhouette matrix.   
     
     
         41 . The apparatus as defined in  claim 40 , wherein the convolution cell selector is to encode the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix. 
     
     
         42 . The apparatus as defined in  claim 36 , wherein the cell retention threshold value includes a value based on at least one of a matrix kernel size or an overlap step size. 
     
     
         43 . A tangible computer-readable medium comprising instructions that, when executed, cause at least one processor to at least:
 generate an encoded silhouette matrix that includes a bitmap area defined by transitions between foreground image cells and background image cells;   reduce a quantity of the foreground image cells within the bitmap area to be processed by convolving the encoded silhouette matrix to generate a convoluted bitmap matrix;   select first cells of the convoluted bitmap matrix that satisfy a cell retention threshold value;   select second cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold value;   generate a contributed blocks matrix by encoding the first cells with respective retention indicators; and   encode the second cells with respective discard indicators, the second cells to be removed from processing by an image classifier.   
     
     
         44 . The computer-readable medium as defined in  claim 43 , wherein the instructions, when executed, cause the at least one processor to generate an eligible block matrix by identifying an outer perimeter based on the respective retention indicators. 
     
     
         45 . The computer-readable medium as defined in  claim 44 , wherein the instructions, when executed, cause the at least one processor to replace cells internal to the outer perimeter with respective discard indicators, the cells internal to the outer perimeter indicative of a portion of the foreground cells. 
     
     
         46 . The computer-readable medium as defined in  claim 45 , wherein the instructions, when executed, cause the at least one processor to mask a detection window with the eligible block matrix. 
     
     
         47 . The computer-readable medium as defined in  claim 43 , wherein the instructions, when executed, cause the at least one processor to:
 convolve a ones matrix kernel to respective cells of the encoded silhouette matrix; and   calculate a sum total of the foreground cells within the ones matrix kernel corresponding to the encoded silhouette matrix.   
     
     
         48 . The computer-readable medium as defined in  claim 47 , wherein the instructions, when executed, cause the at least one processor to encode the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix. 
     
     
         49 . The computer-readable medium as defined in  claim 43 , wherein the instructions, when executed, cause the at least one processor to set the cell retention threshold value based on at least one of a matrix kernel size or an overlap step size. 
     
     
         50 . A system to improve image classification efficiency, the system comprising:
 memory; and   at least one processor to:
 generate an encoded silhouette matrix that includes a bitmap area defined by transitions between foreground image cells and background image cells;
 reduce a quantity of the foreground image cells within the bitmap area to be processed by convolving the encoded silhouette matrix to generate a convoluted bitmap matrix; 
 
 select first cells of the convoluted bitmap matrix that satisfy a cell retention threshold value; 
 select second cells of the convoluted bitmap matrix that do not satisfy the cell retention threshold value; 
 generate a contributed blocks matrix by encoding the first cells with respective retention indicators; and 
 encode the second cells with respective discard indicators, the second cells to be removed from processing by an image classifier. 
   
     
     
         51 . The system as defined in  claim 50 , wherein the at least one processor is to generate an eligible block matrix by identifying an outer perimeter based on the respective retention indicators. 
     
     
         52 . The system as defined in  claim 51 , wherein the at least one processor is to replace cells internal to the outer perimeter with respective discard indicators, the cells internal to the outer perimeter indicative of a portion of the foreground cells. 
     
     
         53 . The system as defined in  claim 52 , wherein the at least one processor is to mask a detection window with the eligible block matrix. 
     
     
         54 . The system as defined in  claim 50 , wherein the at least one processor is to:
 convolve a ones matrix kernel to respective cells of the encoded silhouette matrix; and   calculate a sum total of the foreground cells within the ones matrix kernel corresponding to the encoded silhouette matrix.   
     
     
         55 . The system as defined in  claim 54 , wherein the at least one processor is to encode the convoluted bitmap matrix with the sum total in respective cells of the convoluted bitmap matrix.