iOS开发-图片离线鉴黄 基于TensorFlow nsfw oc版
前言-基于TensorFlow鉴黄
iOS开发中有需要离线鉴黄的地方而且需要oc版本的，找了很久只有swift版本，虽然可以桥接但是还是喜欢用oc的，所以分享一下自己的鉴黄oc版本。
效果
首先要引入Firebase
- 注册firebase
- 注册firebase地址
- 使用自己的项目Bundle ID在firebase上注册
- 生成的 GoogleService-Info.plist 导入自己的项目 (注意和Info.plist同级)
- 使用Cocoapods导入Firebase基础库和鉴黄库
pod 'Firebase/Auth'# login base
pod 'FirebaseMLModelInterpreter', '~> 0.22.0'# 鉴黄
- 在 AppDelegate.m 中
- 引入
#import <Firebase/Firebase.h>
- 在application:didFinishLaunchingWithOptions:函数中写入
[FIRApp configure]; //Firebase 基础配置
鉴黄类的封装
- BaseIdentifyImageManager.h
#import <Foundation/Foundation.h>
#import <UIKit/UIKit.h>
NS_ASSUME_NONNULL_BEGIN
/*
1.pod 'Firebase/Auth'(导入基础库) pod 'FirebaseMLModelInterpreter', '~> 0.22.0'# 鉴黄
2.nsfw.tflite add file
3.只能在真机运行
*/
typedef void(^CompletIdentifyBlock)(BOOL canShow, CGFloat valid);
@interface BaseIdentifyImageManager : NSObject
+ (void)checkImage:(UIImage *)image completBlock:(CompletIdentifyBlock)completBlock;
@end
NS_ASSUME_NONNULL_END
- BaseIdentifyImageManager.m
#import "BaseIdentifyImageManager.h"
#if (TARGET_IPHONE_SIMULATOR && TARGET_OS_IPHONE)
#else //真机
#import <FirebaseMLModelInterpreter/FIRModelInputs.h>
#import <FirebaseMLModelInterpreter/FIRModelOutputs.h>
#import <FirebaseMLModelInterpreter/FIRModelInterpreter.h>
#import <FirebaseMLModelInterpreter/FIRCustomLocalModel.h>
#import <FirebaseMLModelInterpreter/FIRModelInputOutputOptions.h>
static CGFloat const kPersentTag = 0.3;
#endif
@implementation BaseIdentifyImageManager
#pragma mark - manager
+ (void)checkImage:(UIImage *)image completBlock:(CompletIdentifyBlock)completBlock {
#if (TARGET_IPHONE_SIMULATOR && TARGET_OS_IPHONE)
return;
#else //真机
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wdeprecated-declarations"
int INPUTWIDTH = 224;
UIImage *scig = [self _scaleToSize:image size:CGSizeMake(256, 256)];
scig = [self _imageFromImage:scig inRect:CGRectMake(16, 16, INPUTWIDTH, INPUTWIDTH)];
NSString *modelPath = [NSBundle.mainBundle pathForResource:@"nsfw" ofType:@"tflite"];
FIRCustomLocalModel *localModel = [[FIRCustomLocalModel alloc] initWithModelPath:modelPath];
FIRModelInterpreter *fi = [FIRModelInterpreter modelInterpreterForLocalModel:localModel];
CGImageRef cgImage = scig.CGImage;// Your input image
//long imageWidth = CGImageGetWidth(image);
//long imageHeight = CGImageGetHeight(image);
CGContextRef context = [self _createARGBBitmapContextFromImage:cgImage];
CGContextDrawImage(context, CGRectMake(0, 0, INPUTWIDTH, INPUTWIDTH), cgImage);
UInt8 *imageData = CGBitmapContextGetData(context);
CGContextRelease(context);
FIRModelInputs *inputs = [[FIRModelInputs alloc] init];
NSMutableData *inputData = [[NSMutableData alloc] initWithCapacity:0];
int offs =0;
for (int row = offs; row < INPUTWIDTH+offs; row++) {
for (int col = offs; col < INPUTWIDTH+offs; col++) {
long offset = 4 * (row * INPUTWIDTH + col);
//Float32 alph = imageData[offset];
Float32 red = imageData[offset];
Float32 green = imageData[offset+1];
Float32 blue = imageData[offset+2];
Float32 red1 = red - 123;
Float32 green1 = green - 117;
Float32 blue1 = blue - 104;
[inputData appendBytes:&blue1 length:sizeof(blue1)];
[inputData appendBytes:&green1 length:sizeof(green1)];
[inputData appendBytes:&red1 length:sizeof(red1)];
}
}
NSError * error = nil;
[inputs addInput:inputData error:&error];
FIRModelInputOutputOptions *ioOptions = [[FIRModelInputOutputOptions alloc] init];
[ioOptions setInputFormatForIndex:0
type:FIRModelElementTypeFloat32
dimensions:@[@1, @224, @224, @3]
error:&error];
if (error != nil) {
return;
}
[ioOptions setOutputFormatForIndex:0
type:FIRModelElementTypeFloat32
dimensions:@[@1,@2]
error:&error];
if (error != nil) {
return;
}
[fi runWithInputs:inputs options:ioOptions completion:^(FIRModelOutputs *outputs, NSError *error) {
dispatch_async(dispatch_get_main_queue(), ^{
NSArray *ou = [outputs outputAtIndex:0 error:nil];
NSNumber *f = ou[0][0];
float value = 1 - [f floatValue];
if (completBlock) {
BOOL canShow = NO;
if(value >= kPersentTag) {
//色情图片
canShow = YES;
} else {
//不是
canShow = NO;
}
completBlock(canShow, value);
}
//NSLog(@"%@; %@",ou[0][0], ou[0][1]);
//NSLog(@"e:%@",error);
});
}];
#pragma clang diagnostic pop
#endif
}
+ (UIImage *)_scaleToSize:(UIImage *)image size:(CGSize)size {
// 创建一个bitmap的context
// 并把它设置成为当前正在使用的context
UIGraphicsBeginImageContext(size);
// 绘制改变大小的图片
[image drawInRect:CGRectMake(0, 0, size.width, size.height)];
// 从当前context中创建一个改变大小后的图片
UIImage *scaledImage = UIGraphicsGetImageFromCurrentImageContext();
// 使当前的context出堆栈
UIGraphicsEndImageContext();
// 返回新的改变大小后的图片
return scaledImage;
}
+ (UIImage *)_imageFromImage:(UIImage *)image inRect:(CGRect)rect {
//将UIImage转换成CGImageRef
CGImageRef sourceImageRef = [image CGImage];
//按照给定的矩形区域进行剪裁
CGImageRef newImageRef = CGImageCreateWithImageInRect(sourceImageRef, rect);
//将CGImageRef转换成UIImage
UIImage *newImage = [UIImage imageWithCGImage:newImageRef];
//返回剪裁后的图片
return newImage;
}
+ (CGContextRef)_createARGBBitmapContextFromImage:(CGImageRef)inImage {
CGContextRef context = NULL;
CGColorSpaceRef colorSpace;
void * bitmapData;
int bitmapByteCount;
int bitmapBytesPerRow;
// Get image width, height. We'll use the entire image.
size_t pixelsWide = CGImageGetWidth(inImage);
size_t pixelsHigh = CGImageGetHeight(inImage);
// Declare the number of bytes per row. Each pixel in the bitmap in this
// example is represented by 4 bytes; 8 bits each of red, green, blue, and
// alpha.
bitmapBytesPerRow = (int)(pixelsWide * 4);
bitmapByteCount = (int)(bitmapBytesPerRow * pixelsHigh);
// Use the generic RGB color space.
colorSpace = CGColorSpaceCreateDeviceRGB();
if (colorSpace == NULL) {
fprintf(stderr, "Error allocating color space\n");
return NULL;
}
// Allocate memory for image data. This is the destination in memory
// where any drawing to the bitmap context will be rendered.
bitmapData = malloc( bitmapByteCount );
if (bitmapData == NULL) {
fprintf (stderr, "Memory not allocated!");
CGColorSpaceRelease( colorSpace );
return NULL;
}
// Create the bitmap context. We want pre-multiplied ARGB, 8-bits
// per component. Regardless of what the source image format is
// (CMYK, Grayscale, and so on) it will be converted over to the format
// specified here by CGBitmapContextCreate.
CGImageAlphaInfo alpInfo = CGImageGetAlphaInfo(inImage);
context = CGBitmapContextCreate (bitmapData,
pixelsWide,
pixelsHigh,
8, // bits per component
bitmapBytesPerRow,
colorSpace,
alpInfo);
// kCGImageAlphaNoneSkipFirst
if (context == NULL) {
free (bitmapData);
fprintf (stderr, "Context not created!");
}
// Make sure and release colorspace before returning
CGColorSpaceRelease(colorSpace);
return context;
}
#pragma mark - manager end
@end
- 导入训练模型
- ps:在下面分享的Swift版本的鉴黄里面有文件
- 使用
__weak typeof(self) weakSelf = self;
[BaseIdentifyImageManager checkImage:newPhoto completBlock:^(BOOL canShow, CGFloat valid) {
NSLog(@"valid %f", valid);
weakSelf.testView.text = [NSString stringWithFormat:@"图片指数值 - %f", valid];
}];