Closed shilat-soto closed 3 months ago
Most of these directories are created in train_emotions.ipynb. Here is the output of my ls -1 AffectNet: 224 emotion_labels.csv faces full_res full_res_aligned Manually_Annotated_Images orig other_classes training.csv training_filtered.csv training_va.csv validation.csv validation_filtered.csv validation_va.csv
In addition, there is new script to work with recent format of AffectNet dataset in train_affectnet_march2021_pytorch.ipynb
Thank you very much for the quick reply. The project I downloaded from GitHub does not have these files What do they contain? are they needed for a model for facial emotion recognition? The model looks really suitable for me and I would love to use it, I just want to figure out how to use it as best as possible
Most of these folders is created by my script, as I told before. The original files and folder are disctibuted by the authors of AffectNet, and I cannot share their data. However, if you do not need to train your own models with my code, you could just use my models. For simplicity, I creted two libraries, hsemotion that works with PyTorch models and hsemotion-onnx that needs ONNX. These repositories contains demos in their demo folder that you could use to work wit my models
Closing due to inactivity
What additional files need to be created in the project and what is their content I would be happy to list a file tree for the project files For example, the files behind the code: AFFECT_DATA_DIR=ALL_DATA_DIR#+'AffectNet/' AFFECT_TRAIN_DATA_DIR = AFFECT_DATA_DIR+'full_res/train' AFFECT_VAL_DATA_DIR = AFFECT_DATA_DIR+'full_res/val' AFFECT_IMG_TRAIN_DATA_DIR = AFFECT_DATA_DIR+str(INPUT_SIZE[0])+'/train' AFFECT_IMG_VAL_DATA_DIR = AFFECT_DATA_DIR+str(INPUT_SIZE[0])+'/val' AFFECT_TRAIN_ORIG_DATA_DIR = AFFECT_DATA_DIR+'orig/train' AFFECT_VAL_ORIG_DATA_DIR = AFFECT_DATA_DIR+'orig/val'