https://github.com/hariprasath-v/zindi_cgiar_crop_damage_classification_challenge

Crop damage classification
https://github.com/hariprasath-v/zindi_cgiar_crop_damage_classification_challenge

computer-vision cv2 exploratory-data-analysis image-classification imagehash logloss machine-learning pandas pytorch sklearn wandb zindi

Last synced: 8 months ago
JSON representation

Crop damage classification

• Host: GitHub
• URL: https://github.com/hariprasath-v/zindi_cgiar_crop_damage_classification_challenge
• Owner: hariprasath-v
• License: apache-2.0
• Created: 2024-03-06T15:31:46.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-03-06T15:54:54.000Z (over 1 year ago)
• Last Synced: 2025-01-13T01:44:55.149Z (9 months ago)
• Topics: computer-vision, cv2, exploratory-data-analysis, image-classification, imagehash, logloss, machine-learning, pandas, pytorch, sklearn, wandb, zindi
• Homepage:
• Size: 6.84 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Zindi_Cgiar_crop_damage_classification_challenge

### Competition hosted on Zindi

# About

### Create a machine-learning algorithm to classify crops into categories: Good growth (G), Drought (DR), Nutrient Deficient (ND), Weed (WD), and Other (including pest, disease or wind damage). The data for this challenge is a collection of smartphone images of crops.

### The Final Competition score is 0.696384917

### Final Leaderboard Rank is 152.

### The Evaluation Metric is  Log Loss.

### File information

 

 * EDA [![Open in Kaggle](https://img.shields.io/static/v1?label=&message=Open%20in%20Kaggle&labelColor=grey&color=blue&logo=kaggle)](https://www.kaggle.com/code/hari141v/cgiar-crop-damage-classification-challenge-eda)

    #### Basic image information analysis

    #### Images RGB color analysis

    #### Image similarity analysis

    #### Packages Used,

        * seaborn 

        * Pandas

        * Numpy

        * Matplotlib

        * imagehash

        * distance

        * Image

        * cv2

* Model

  ### Trained ViT Base 16 Patch 224 model on five-fold training data with various augmentations. Ten epochs were used to train the five-fold dataset, and early stopping was implemented to control overfitting by monitoring the validation log loss. The test data was predicted using the five-fold model, and test-time augmentation was applied to ensure confident predictions. The model's performance was tracked using WANDB.

  ### [Model Log](https://wandb.ai/hari141v/Cgiar_Crop_Damage_Classification_10)