RetinaFace Post-Processing is a module contains code for post-processing of RetinaFace inference results
Project description
RetinaFace Post-Processing
Module for post-processing of RetinaFaceNet inference results.
Model outputs are floating points tensors:
-
name:
face_rpn_cls_prob
, shape:1, 16800, 2
, format:B, A*C, 2
, represents detection scores for 2 classes: background and face. -
name:
face_rpn_bbox_pred
, shape:1, 16800, 4
, format:B, A*C, 4
, represents detection box deltas. -
name:
face_rpn_landmark_pred
, shape:1, 16800, 10
, format:B, A*C, 10
, represents facial landmarks.
For each output format:
B
- batch sizeA
- number of anchorsC
- sum of products of dimensions for each stride,C = H32 * W32 + H16 * W16 + H8 * W8
H
- feature height with the corresponding strideW
- feature width with the corresponding stride
Detection box deltas have format [dx, dy, dh, dw]
, where:
(dx, dy)
- regression for center of bounding box(dh, dw)
- regression by height and width of bounding box
Facial landmarks have format [x1, y1, x2, y2, x3, y3, x4, y4, x5, y5]
, where:
(x1, y1)
- coordinates of left eye(x2, y2)
- coordinates of rights eye(x3, y3)
- coordinates of nose(x4, y4)
- coordinates of left mouth corner(x5, y5)
- coordinates of right mouth corner
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for retinaface_post_processing-0.0.3.tar.gz
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9faec4952255ca4816633ec53ba159e63698bd63815c6a8598472b57e486fe55 |
|
MD5 | c9ae5210ee9dcab5ca71179d0692b8ec |
|
BLAKE2b-256 | f08efb6dd4764d45cc544a82c4a2a67ac01339ee05ac11a4a61a677a1e1f7077 |
Hashes for retinaface_post_processing-0.0.3-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 6bc7701bcd87aaa8a94994c320b4da2a80177e52d9892f17baf0de494baa2cd9 |
|
MD5 | 729b68cdbdbd105353dfe6be7f77bebd |
|
BLAKE2b-256 | 33a48a41f14f73166404737c684620fd3eb7cd936631d19c3d92870a368fb174 |