return the max value in rows(d=0) or in cols(d=1) arr = [4 9 5 7 8 5]
return the max value in rows(d=0) or in cols(d=1) arr = [4 9 5 7 8 5]
argmax2(arr, 1) will return 3, 1 argmax2(arr, 2) will return 2, 2, 1
todo: this maybe removed
Get the overlap between boxes and query_boxes
Get the overlap between boxes and query_boxes
(N, 4) ndarray of float
(K, >=4) ndarray of float
overlaps: (N, K) ndarray of overlap between boxes and query_boxes
Note that the output are stored in input deltas
Note that the output are stored in input deltas
(N, 4)
(N, 4a)
BBox vote result
BBox vote result
N
N * 4
M
M * 4
Clip boxes to image boundaries.
Clip boxes to image boundaries. set the score of all boxes with any side smaller than minSize to 0
N * 4a
height of image
width of image
min height limit
min width limit
scores for boxes
the number of boxes kept (score > 0)
decode batch
Encode BBox
get the areas of boxes
get the areas of boxes
N * 4 tensor
buffer to store the results
areas array
Bounding-box regression targets (bboxTargetData) are stored in a compact form N x (class, tx, ty, tw, th) * This function expands those targets into the 4-of-4*K representation used by the network (i.e.
Bounding-box regression targets (bboxTargetData) are stored in a compact form N x (class, tx, ty, tw, th) * This function expands those targets into the 4-of-4*K representation used by the network (i.e. only one class has non-zero targets). * Returns: bbox_target (ndarray): N x 4K blob of regression targets bbox_inside_weights (ndarray): N x 4K blob of loss weights
Get Confidence scores
Get groud truth indices from result
Get ground truth from result
Get location prediction result
Concat multiple 2D tensors horizontally
Project bbox onto the coordinate system defined by src_bbox.
Select tensor from matrix
Select tensor from matrix
source matrix
indices array
dimension
indice length
based tensor
selected tensor
Select tensor from matrix
Select tensor from matrix
source matrix
indices array
dimension
indice length
based tensor
selected tensor
update with 2d tensor, the range must be equal to the src tensor size
Concat multiple 1D tensors vertically
Concat multiple 2D tensors vertically
Concat multiple 2D tensors vertically
tensor list