This repository contains scripts for the Deep Learning Challenge #1 hosted by hackersEarth.
The problem statement, dataset and data description can be found here. https://www.hackerearth.com/challenge/competitive/deep-learning-challenge-1/machine-learning/identify-the-objects/
Approach:
Used the model trained on VGG16
Result
Rank: 42 on 4500+ participants (top 1 percentile)
Folder Structure
Root Folder
|
|
|_____ Data
|
|| Test_img
|
|| Train_img
|
|| Test and Train Image Names in csv Files (test.csv and train.csv)
|
|
|__ Submissions
|
|
|_____ Trained Model (Not Used Currently)
|
|
|_____ HackersEarth_Submission (Code)
|
|| Python code as Files (keras_vggmodel.ipynb, keras_vggmodel_v1.ipynb and README)
|
|
|__ README (File)
Problem: Predict the product category
Duration: 1st September 2017 to 30th September 2017
Prize: USD 700, USD 500
Evaluation Metric: F1-Score