metadata
license: cc-by-nc-4.0
base_model: MCG-NJU/videomae-base
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: videomae-base-finetuned-fight-nofight-subset2
results: []
datasets:
- Pinwheel/ActsOfAgression
videomae-base-finetuned-fight-nofight-subset2
NOTE: This is experimentational if youre expecting this to work accurately (it wont) or be useful should probably look eslewhere😛
This model is a fine-tuned version of MCG-NJU/videomae-base on the Acts of Agression (cttv footage fights) dataset. It achieves the following results on the evaluation set:
- Loss: 0.5190
- Accuracy: 0.7435
Model description
Classifies video input into "Fight" or "No Fight" Class
Intended uses & limitations
Can be used to detect fights/crime in cctv footage
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 252
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.5145 | 0.25 | 64 | 0.7845 | 0.5075 |
| 0.607 | 1.25 | 128 | 0.6886 | 0.6343 |
| 0.3986 | 2.25 | 192 | 0.5106 | 0.7463 |
| 0.3632 | 3.24 | 252 | 0.7408 | 0.6716 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2
- Datasets 2.1.0
- Tokenizers 0.15.1