Skip to content

human_face_recognition

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

human_face_recognition

README.md

Human Face Recognition Models

Model list

Chip MFN_S8_V1 MBF_S8_V1
ESP32-S3 alt text alt text
ESP32-P4 alt text alt text

Model Latency

name input(hwc) preprocess(ms) model(ms) postprocess(ms)
mfn_s8_v1_s3 112 * 112 * 3 5.6 248.8 0.1
mfn_s8_v1_p4 112 * 112 * 3 2.9 93.0 0.1
mbf_s8_v1_s3 112 * 112 * 3 5.6 1072.4 0.1
mbf_s8_v1_p4 112 * 112 * 3 2.9 188.2 0.1

Model Metrics

Model Params(M) GFLOPs TAR@FAR=1E-4 on IJB-C(%)
mfn_s8_v1 1.2 0.46 90.03
mbf_s8_v1 3.4 0.90 93.94

Model Usage

HumanFaceFeat accepts a HumanFaceFeat::model_type_t parameter. It has a default value determined by default model option in menuconfig.

How to New HumanFaceFeat

Only One Model

HumanFaceFeat *feat = new HumanFaceFeat();

Multiple Models

// use MFN_S8_V1
HumanFaceFeat *feat = new HumanFaceFeat(HumanFaceFeat::MFN_S8_V1);
// use MBF_S8_V1
// HumanFaceFeat *feat = new HumanFaceFeat(HumanFaceFeat::MBF_S8_V1);

Note

If multiple models is flashed or stored in sdcard, in addition to the default model, you can pass an explicit parameter to HumanFaceFeat to use one of them.

Configurable Options in Menuconfig

See Kconfig.

Model to Flash

  • CONFIG_FLASH_HUMAN_FACE_FEAT_MFN_S8_V1
  • CONFIG_FLASH_HUMAN_FACE_FEAT_MBF_S8_V1

Whether to flash the model when model location is set to FLASH rodata or FLASH partition.

Default Model

  • CONFIG_HUMAN_FACE_FEAT_MFN_S8_V1
  • CONFIG_HUMAN_FACE_FEAT_MBF_S8_V1

Default model to use if no parameter is passed to HumanFaceFeat.

Model Location

  • CONFIG_HUMAN_FACE_FEAT_MODEL_IN_FLASH_RODATA
  • CONFIG_HUMAN_FACE_FEAT_MODEL_IN_FLASH_PARTITION
  • CONFIG_HUMAN_FACE_FEAT_MODEL_IN_SDCARD

This component supports to load model from three different locations.

Note

If model location is set to FLASH partition, partition.csv must contain a partition named human_face_feat, and the partition should be big enough to hold the model file.

SDCard Directory

  • CONFIG_HUMAN_FACE_FEAT_MODEL_SDCARD_DIR

When model locates in sdcard, you can change the model directory relative to the sdcard mount point.

The default value of this option is models/s3 for ESP32S3 and models/p4 for ESP32P4. When using default value, just copy models folder to sdcard root directory.

Note

Do not change the model name when copy the models to sdcard.

Human Face Recognizer

This component also contains HumanFaceRecognizer. It's a integration of HumanFaceFeat and dl::recognition::DataBase. If you want to enroll/recognize a human face, in addition to HumanFaceRecognizer, you also need HumanFaceDetect. See How to New HumanFaceDetect.

How to New Human Face Recognizer

HumanFaceRecognizer *human_face_recognizer = new HumanFaceRecognizer("path/to/database");

How to Enroll a Human Face

dl::image::img_t img = {.data=DATA, .width=WIDTH, .height=HEIGHT, .pix_type=PIX_TYPE};
human_face_recognizer->enroll(img, human_face_detect->run(img));

More details, see dl::image::img_t.

How to Recognize a Human Face

dl::image::img_t img = {.data=DATA, .width=WIDTH, .height=HEIGHT, .pix_type=PIX_TYPE};
std::vector<dl::recognition::result_t> res = human_face_recognizer->recognize(img, human_face_detect->run(img));

More details, see dl::image::img_t and dl::recognition::result_t.

How to Delete Feature in Database

Delete all Features

human_face_recognizer->clear_all_feats();

Delete Last Feature

human_face_recognizer->delete_last_feat();

Delete Feature by Index

human_face_recognizer->delete_feat(index);