CNN face detector (deep learning face detection) is modeled as PHP class. Currently, it only has constructor (which loads a model) and detect() method. I tried to make it resamble to Python implementation. Similar to Python, also added support for upsampling. Returned value is array with numbers as keys where number of keys is equal to number of found faced. Each element is associative array that has "top", "left", "right", "bottom" and "detection_confidence" keys. All errors are propagated using exceptions. Added two new error tests. I didn't want to rely on presence of model, so tests are not having great coverage. If we are OK to download models in tests, it would allow us to have far better coverage. What is missing: * Testing with models (should we test that?) * detect_mult method (similar to what Python have; however, I consider that this should not block pushing this change)pull/1/head
Branko Kokanovic
6 years ago
parent
d12d75c69e
commit
f8cc0e48d3
@ -0,0 +1,108 @@
|
||||
#include "../php_pdlib.h"
|
||||
#include "cnn_face_detection.h"
|
||||
|
||||
#include <zend_exceptions.h>
|
||||
#include <dlib/image_processing/frontal_face_detector.h>
|
||||
#include <dlib/gui_widgets.h>
|
||||
#include <dlib/image_io.h>
|
||||
#include <dlib/dnn.h>
|
||||
#include <iostream>
|
||||
|
||||
using namespace dlib;
|
||||
using namespace std;
|
||||
|
||||
static inline cnn_face_detection *php_cnn_face_detection_from_obj(zend_object *obj) {
|
||||
return (cnn_face_detection*)((char*)(obj) - XtOffsetOf(cnn_face_detection, std));
|
||||
}
|
||||
|
||||
#define Z_CNN_FACE_DETECTION_P(zv) php_cnn_face_detection_from_obj(Z_OBJ_P((zv)))
|
||||
|
||||
PHP_METHOD(CnnFaceDetection, __construct)
|
||||
{
|
||||
char *sz_cnn_face_detection_model_path;
|
||||
size_t cnn_face_detection_model_path_len;
|
||||
|
||||
cnn_face_detection *cfd = Z_CNN_FACE_DETECTION_P(getThis());
|
||||
|
||||
if (NULL == cfd) {
|
||||
php_error_docref(NULL TSRMLS_CC, E_ERROR, "Unable to find obj in CnnFaceDetection::__construct()");
|
||||
return;
|
||||
}
|
||||
|
||||
if (zend_parse_parameters(ZEND_NUM_ARGS(), "s",
|
||||
&sz_cnn_face_detection_model_path, &cnn_face_detection_model_path_len) == FAILURE){
|
||||
zend_throw_exception_ex(zend_ce_exception, 0 TSRMLS_CC, "Unable to parse face_detection_model_path");
|
||||
return;
|
||||
}
|
||||
|
||||
try {
|
||||
string cnn_face_detection_model_path(
|
||||
sz_cnn_face_detection_model_path, cnn_face_detection_model_path_len);
|
||||
net_type *pnet = new net_type;
|
||||
deserialize(cnn_face_detection_model_path) >> *pnet;
|
||||
cfd->net = pnet;
|
||||
} catch (exception& e) {
|
||||
zend_throw_exception_ex(zend_ce_exception, 0 TSRMLS_CC, e.what());
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
PHP_METHOD(CnnFaceDetection, detect)
|
||||
{
|
||||
char *img_path;
|
||||
size_t img_path_len;
|
||||
long upsample_num = 1;
|
||||
|
||||
if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|l", &img_path, &img_path_len, &upsample_num) == FAILURE){
|
||||
zend_throw_exception_ex(zend_ce_exception, 0 TSRMLS_CC, "Unable to parse detect arguments");
|
||||
RETURN_FALSE;
|
||||
}
|
||||
|
||||
try {
|
||||
cnn_face_detection *cfd = Z_CNN_FACE_DETECTION_P(getThis());
|
||||
|
||||
pyramid_down<2> pyr;
|
||||
matrix<rgb_pixel> img;
|
||||
load_image(img, img_path);
|
||||
|
||||
// Upsampling the image will allow us to detect smaller faces but will cause the
|
||||
// program to use more RAM and run longer.
|
||||
//
|
||||
unsigned int levels = upsample_num;
|
||||
while (levels > 0)
|
||||
{
|
||||
levels--;
|
||||
pyramid_up(img, pyr);
|
||||
}
|
||||
|
||||
net_type *pnet = cfd->net;
|
||||
auto dets = (*pnet)(img);
|
||||
int rect_count = 0;
|
||||
array_init(return_value);
|
||||
|
||||
// Scale the detection locations back to the original image size
|
||||
// if the image was upscaled.
|
||||
//
|
||||
for (auto&& d: dets) {
|
||||
d.rect = pyr.rect_down(d.rect, upsample_num);
|
||||
// Create new assoc array with dimensions of found rectt and confidence
|
||||
//
|
||||
zval rect_arr;
|
||||
array_init(&rect_arr);
|
||||
add_assoc_long(&rect_arr, "left", d.rect.left());
|
||||
add_assoc_long(&rect_arr, "top", d.rect.top());
|
||||
add_assoc_long(&rect_arr, "right", d.rect.right());
|
||||
add_assoc_long(&rect_arr, "bottom", d.rect.bottom());
|
||||
add_assoc_double(&rect_arr, "detection_confidence", d.detection_confidence);
|
||||
// Add this assoc array to returned array
|
||||
//
|
||||
add_next_index_zval(return_value, &rect_arr);
|
||||
}
|
||||
}
|
||||
catch (exception& e)
|
||||
{
|
||||
zend_throw_exception_ex(zend_ce_exception, 0 TSRMLS_CC, e.what());
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
@ -0,0 +1,36 @@
|
||||
//
|
||||
// Created by branko at kokanovic dot org on 2018/7/16.
|
||||
//
|
||||
|
||||
#ifndef PHP_DLIB_CNN_FACE_DETECTION_H
|
||||
#define PHP_DLIB_CNN_FACE_DETECTION_H
|
||||
|
||||
#include <dlib/dnn.h>
|
||||
|
||||
using namespace dlib;
|
||||
|
||||
template <long num_filters, typename SUBNET> using con5d = con<num_filters,5,5,2,2,SUBNET>;
|
||||
template <long num_filters, typename SUBNET> using con5 = con<num_filters,5,5,1,1,SUBNET>;
|
||||
|
||||
template <typename SUBNET> using downsampler = relu<affine<con5d<32, relu<affine<con5d<32, relu<affine<con5d<16,SUBNET>>>>>>>>>;
|
||||
template <typename SUBNET> using rcon5 = relu<affine<con5<45,SUBNET>>>;
|
||||
|
||||
using net_type = loss_mmod<con<1,9,9,1,1,rcon5<rcon5<rcon5<downsampler<input_rgb_image_pyramid<pyramid_down<6>>>>>>>>;
|
||||
|
||||
typedef struct _cnn_face_detection {
|
||||
net_type *net;
|
||||
zend_object std;
|
||||
} cnn_face_detection;
|
||||
|
||||
ZEND_BEGIN_ARG_INFO_EX(cnn_face_detection_ctor_arginfo, 0, 0, 1)
|
||||
ZEND_ARG_INFO(0, cnn_face_detection_model_path)
|
||||
ZEND_END_ARG_INFO()
|
||||
PHP_METHOD(CnnFaceDetection, __construct);
|
||||
|
||||
ZEND_BEGIN_ARG_INFO_EX(cnn_face_detection_detect_arginfo, 0, 0, 2)
|
||||
ZEND_ARG_INFO(0, img_path)
|
||||
ZEND_ARG_INFO(0, upsample_num)
|
||||
ZEND_END_ARG_INFO()
|
||||
PHP_METHOD(CnnFaceDetection, detect);
|
||||
|
||||
#endif //PHP_DLIB_CNN_FACE_DETECTION_H
|
||||
@ -0,0 +1,15 @@
|
||||
--TEST--
|
||||
Testing CnnFaceDetection constructor without arguments
|
||||
--SKIPIF--
|
||||
<?php if (!extension_loaded("pdlib")) print "skip"; ?>
|
||||
--FILE--
|
||||
<?php
|
||||
try {
|
||||
new CnnFaceDetection();
|
||||
} catch (Exception $e) {
|
||||
var_dump($e->getMessage());
|
||||
}
|
||||
?>
|
||||
--EXPECT--
|
||||
Warning: CnnFaceDetection::__construct() expects exactly 1 parameter, 0 given in /home/branko/pdlib/tests/cnn_face_detection_ctor_error.php on line 3
|
||||
string(41) "Unable to parse face_detection_model_path"
|
||||
@ -0,0 +1,14 @@
|
||||
--TEST--
|
||||
Testing CnnFaceDetection constructor with model that do not exist
|
||||
--SKIPIF--
|
||||
<?php if (!extension_loaded("pdlib")) print "skip"; ?>
|
||||
--FILE--
|
||||
<?php
|
||||
try {
|
||||
new CnnFaceDetection("foo");
|
||||
} catch (Exception $e) {
|
||||
var_dump($e->getMessage());
|
||||
}
|
||||
?>
|
||||
--EXPECT--
|
||||
string(31) "Unable to open foo for reading."
|
||||
Loading…
Reference in new issue