insightface/iccv19-challenge

The Lightweight Face Recognition Challenge & Workshop will be held in conjunction with the International Conference on Computer Vision (ICCV) 2019, Seoul Korea.

Firstly please read the workshop homepage carefully about the training and testing rules.

Test Submission Server[TODO]

==================

How To Start:

Training:

1. Download ms1m-retinaface from baiducloud or dropbox and unzip it to $INSIGHTFACE_ROOT/datasets/
2. Go into $INSIGHTFACE_ROOT/recognition/
3. Refer to the retina dataset config section in sample_config.py and copy it to your ownconfig.py.
4. Start training with CUDA_VISIBLE_DEVICES='0,1,2,3' python -u train.py --dataset retina --network [your-network] --loss arcface. It will output the accuracy of lfw, cfp_fp and agedb_30 every 2000 batches by default.

---

Testing:

1. testdata-image from baiducloud or dropbox. These face images are all pre-processed and aligned so no need to do further modification.
2. To download testdata-video from iQIYI, please visit http://challenge.ai.iqiyi.com/data-cluster. You must download iQIYI-VID-FACE.z01, iQIYI-VID-FACE.z02 and iQIYI-VID-FACE.zip after signin. These face images are all pre-processed and aligned so no need to do further modification.
   1. To unzip: zip iQIYI_VID_FACE.zip -s=0 --out iQIYI_VID_FACE_ALL.zip; unzip iQIYI_VID_FACE_ALL.zip
   2. We can get a directory named iQIYI_VID_FACE after decompression. Then we have to move video_filelist.txt in testdata-image package to iQIYI_VID_FACE/filelist.txt, to indicate the order of videos in our submission feature files.
3. To generate image feature file: check gen_image_feature.py
4. To generate video feature file: check gen_video_feature.py
5. Submit feature binary file to the right section on test server.

You can also check the verification performance during training time on LFW,CFP_FP,AgeDB_30 datasets.

---

Evaluation:

Final ranking is determined by accuracy only, for all valid submissions. For example, score of track-1 will be calculated by TAR_glint-light+TAR_iqiyi-light while TAR_glint-large+TAR_iqiyi-large for track-2.

---

Discussion:

https://github.com/deepinsight/insightface/issues/632

---

Baseline:

1. Network y2(a deeper mobilefacenet): 933M FLOPs. TAR_image: [TODO], TAR_video: [TODO]
2. Network r100fc(ResNet100FC-IR): 24G FLOPs. TAR_image: [TODO], TAR_video: [TODO]

---

Candidate solutions:

1. Use slightly deeper or wider mobile-level networks.
2. Try different training methods/losses than straightforward arcface.
3. OctConv, to reduce FLOPs.
4. HRNet, for large FLOPs track.