🛠 Contribute: Found a typo? Or any other change in the description that you would like to see? Please consider sending us a pull request in the public repo of the challenge here.
You want to sign up for a new exciting web service and you’re almost done but the pesky CAPTCHA form has popped up and is now delaying you. If you have been annoyed by the CAPTCHA we feel you! This challenge is to get even -- create your own ML model that will identify values correctly and you may never have to solve a CAPTCHA form again.
For this puzzle, you will be given a dataset of CAPTCHA images created specifically for this challenge. The training set consists of two columns [filename] and [label] . Your task is to predict the captchas with the proper case and save them in a csv with labels filename and label for the test set.
Understand with code! Here is
getting started code for you.
The CAPCHA dataset is a in-house created dataset designed specifically for this challenge. The train set and test sets consist of images of various types of captcha. In the train set you are also given a
train.csv which has two columns
label. The captchas are a mix of capital letters, lowercase letters and numbers; representative of a real life captcha.
You are to predict the captchas with the proper case and save them in a csv with labels
label for the test set.
Following files are available in the
10,000samples) This csv file contains the labels for each of the images in the train.zip .
10,000samples) This zip file contatins all the train images on which you can train your model on.
5,000samples) This csv file contains the filename for each of the images in the test.zip for which labels needs to be predicted.
5,000samples) The tar file contains all the test images, on which the actual evaluation will take place.
- Prepare a CSV containing
labeland predicted the string that is in the captcha with proper case.
- Name of the above file should be submission.csv.
- Sample submission format available at sample_submission.csv in the resorces section.
Make your first submission here 🚀 !!
🖊 Evaluation Criteria
During evaluation mean over normalised Levenshtein Similarity Score will be used to test the efficiency of the model.
The score of only 60% of the test data will be revealed during the competition.
- 💪 Challenge Page: https://www.aicrowd.com/challenges/CPTCHA
- 🗣️ Discussion Forum: https://www.aicrowd.com/challenges/CPTCHA/discussion
- 🏆 Leaderboard: https://www.aicrowd.com/challenges/CPTCHA/leaderboards