visionjo / facerec-bias-bfw Goto Github PK

Hi,

It would be great if you could clarify a few questions regarding this dataset please.

1. Is it possible for you to provide the MTCNN output face detections (bounding boxes and facial landmarks) for the face samples in BFW?

2. Am I right in assuming MTCNN takes as input the images in "face-samples" folder of the dataset? If yes, what settings do we use with MTCNN in order for us to detect a face correctly on all the facial images provided in face-samples? If not, can you help us reproduce your face detection results by providing us with the original images on which the MTCNN was run to obtain the results in face-samples?

3. Are the images in facial-samples actually crops which are aligned?

Thanks in advance for your help.

Setup the project for sphinx.

Include clear instruction on how to maintain (i.e., once in place, we'll include as part of the build process (see in docs/)

Setup for tutorials on the different concepts and experiments done as part of this line of work (i.e., facial bias and BFW database)

Hi,
Thanks for your great work and sharing of the code on these two papers !
It takes me days to read the paper and go through the repository and I have a few questions:

(2)
Do you have the code for training the features (asian_females, asian_males, black_females, black_males, indian_females, indian_males,...). Since I have a hard time finding something like train.py (e.g. the loss function and training process).
(I suppose the released code is mainly on image pre-processing and result analysis)
(Since BFW dataset is not as large as other face dataset and it may possible for me to train it from scratch on one GPU)

(3)
I am little confused about how the BFW is used in two papers, as I understand:

in paper Face Recognition: Too Bias, or Not Too Bias? , the train and test model are as follows:
train: CASIA_webface trained using Sphereface loss
test: LFW
where does BFW dataset not used in training in this set of experiments?

in paper Balancing Biases and Preserving Privacy on Balanced Faces in the Wild the train, test model are as follows:
tain:
(1) MS1M trained using Arcface loss --> to get 512-dim embedding (f_in in Fig.6)
(2) BFW dataset is used to train the encoder and two classifiers in Fig 6
test: 4-folds used for training and 1-fold used for testing (using the best threshold chosen)

is that right?

(4)
There are some difference from "bfw-v0.1.5-datatable.csv" and the TABLE-2 in paper 2:
for example: there are 921379 records in TABLE-2 while ther are 923898 records from the csv file?
and there is no "{dir_meta}thresholds.pkl" file.

Thanks for your time and any help would be appreciated !

Design pipeline to validate age labels inferred by pre-trained classify.

Split into coarse groups (e.g., 3-- young, middle, old)

Hey,

Thanks for the awesome work!

I wanted to know how I can modify the repo to use for face identification task instead of verification.

Any help would be highly appreciated.

3 Figures based on the paper "Face Recognition: Too Bias, or Not Too Bias" are

1. DET curves: FPR versus FNR by moving threshold
2. Score distributions for genuine and imposter using violin plots
3. Confusion matrix for Rank 1 and any Rank.

Add tests and to tests as seen fit (i.e., if something goes wrong, and should be included as part of test or assertion should be added, please do so).

ADDED:
Unit tests, assertions, such, along with the schema to be followed as part of the workflow in this project from there onward. (i.e., in future to set up right off the bat; get a system / template in place)

Project plan (lead: Dylan; support: Rohan):

• what features to include
• Specifications
• Interface layout (use lucidchart or equivalent)
• Division of tasks and proposed timeline

design report (layout, contents, etc.)

e.g., https://www.lucidchart.com/pages/templates/wireframe/high-fidelity-wireframe-template

The root README currently contains all figures in paper-- move this to results/README. For root README, just share an overview and motivational figure of paper. More focus on project usage, contents, such