jhoppanne's picture
Update README.md
43326bd verified
metadata
license: apache-2.0
base_model: microsoft/resnet-152
tags:
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
model-index:
  - name: Dogs-Breed-Image-Classification-V2
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: imagefolder
          type: imagefolder
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8408163265306122

Dogs-Breed-Image-Classification-V2

This model is a fine-tuned version of microsoft/resnet-152 on the Standford dogs dataset. It achieves the following results on the evaluation set:

  • Loss: 1.0115
  • Accuracy: 0.8408

Model description

This model was trained using dataset from Kaggle - Standford dogs dataset

Quotes from the website: The Stanford Dogs dataset contains images of 120 breeds of dogs from around the world. This dataset has been built using images and annotation from ImageNet for the task of fine-grained image categorization. It was originally collected for fine-grain image categorization, a challenging problem as certain dog breeds have near identical features or differ in colour and age.

citation: Aditya Khosla, Nityananda Jayadevaprakash, Bangpeng Yao and Li Fei-Fei. Novel dataset for Fine-Grained Image Categorization. First Workshop on Fine-Grained Visual Categorization (FGVC), IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2011. [pdf] [poster] [BibTex]

Secondary: J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, ImageNet: A Large-Scale Hierarchical Image Database. IEEE Computer Vision and Pattern Recognition (CVPR), 2009. [pdf] [BibTex]

Intended uses & limitations

This model is fined tune solely for classifiying 120 species of dogs.

Training and evaluation data

75% training data, 25% testing data.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-06
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 483 4.6525 0.7382
4.7329 2.0 966 4.3558 0.7298
4.5033 3.0 1449 3.9568 0.7471
4.1405 4.0 1932 3.5160 0.7782
3.7176 5.0 2415 3.0805 0.7946
3.293 6.0 2898 2.6907 0.8021
2.8898 7.0 3381 2.3044 0.8126
2.5343 8.0 3864 2.0091 0.8177
2.2188 9.0 4347 1.7910 0.8126
1.9698 10.0 4830 1.6015 0.8194
1.7532 11.0 5313 1.4383 0.8220
1.586 12.0 5796 1.3355 0.8264
1.4533 13.0 6279 1.2467 0.8260
1.336 14.0 6762 1.1575 0.8313
1.2641 15.0 7245 1.1038 0.8321
1.185 16.0 7728 1.0606 0.8395
1.1329 17.0 8211 1.0178 0.8398
1.0977 18.0 8694 1.0115 0.8408
1.0732 19.0 9177 0.9945 0.8381
1.0508 20.0 9660 0.9930 0.8393

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.3.0
  • Datasets 2.15.0
  • Tokenizers 0.15.1