https://github.com/macaodha/inat_comp_2018

CNN training code for iNaturalist 2018 image classification competition
https://github.com/macaodha/inat_comp_2018

Last synced: 8 days ago
JSON representation

CNN training code for iNaturalist 2018 image classification competition

• Host: GitHub
• URL: https://github.com/macaodha/inat_comp_2018
• Owner: macaodha
• Created: 2018-02-12T21:49:47.000Z (almost 7 years ago)
• Default Branch: master
• Last Pushed: 2023-08-10T15:43:00.000Z (over 1 year ago)
• Last Synced: 2024-08-01T15:30:14.640Z (3 months ago)
• Language: Python
• Size: 9.77 KB
• Stars: 82
• Watchers: 8
• Forks: 23
• Open Issues: 0
• Metadata Files:
  • Readme: readme.md

Awesome Lists containing this project

README

        
# iNaturalist Competition 2018 Training Code  

This code finetunes an Inception V3 model (pretrained on ImageNet) on the iNaturalist 2018 competition [dataset](https://github.com/visipedia/inat_comp).

### Training

The network was trained on Ubuntu 16.04 using PyTorch 0.3.0. Each training epoch took about 1.5 hours using a GTX Titan X.  

The links for the raw data are available [here](https://github.com/visipedia/inat_comp).

We also provide a trained model that can be downloaded from [here](https://data.caltech.edu/records/7e9te-sae53/files/iNat_2018_InceptionV3.pth.tar).

Every epoch the code will save a checkpoint and the current best model according to validation accuracy.  

Training for 75 epochs results in a top one accuracy of 60.20% and top three of 77.91% on the validation set.

### Ideas for Improvement  

* Train/test on higher resolution images.  

* Make use of the taxonomy at training time (already included in data loader).  

* Address long tail distribution.

### Submission File

By setting the following flags it's possible to generate a submission file for the competition.

```python

    evaluate = True

    save_preds = True

    resume = 'model_path/iNat_2018_InceptionV3.pth.tar'  # path to trained model

    val_file = 'ann_path/test2018.json'                  # path to test file

    data_root = 'data_path/inat2018/images/'             # path to test images

    op_file_name = 'inat2018_test_preds.csv'             # submission filename

```