Recognition of traffic signs is a challenging real-world problem of high industrial relevance. Although commercial systems have reached the market and several studies on this topic have been published, systematic unbiased comparisons of different approaches are missing and comprehensive benchmark datasets are not freely available.
\n\n\nTraffic sign recognition is a multi-class classification problem with unbalanced class frequencies. Traffic signs can provide a wide range of variations between classes in terms of color, shape, and the presence of pictograms or text. However, there exist subsets of classes (e. g., speed limit signs) that are very similar to each other.
\n\n\nThe classifier has to cope with large variations in visual appearances due to illumination changes, partial occlusions, rotations, weather conditions, etc.
\n\n\nHumans are capable of recognizing the large variety of existing road signs with close to 100% correctness. This does not only apply to real-world driving, which provides both context and multiple views of a single traffic sign, but also to the recognition from single images.
\n\nIt's a great dataset for multi-class image classification problem.
\n\nUpdate: If you have downloaded the data before 4/14/2020, please re-download the data. The labels in the new dataset are now one-hot encoded
\nThis is the Kaggle mini-competition, part of CSCI 4622 that you will be participating. In this competition, you will work with a dataset to classify German traffic signs. This is a simple clean dataset in which each image is 32x32x3 sized which is flattened for convenience purpose to 3072 features. Training and validation data are available in the 'Data' tab.
\n\n\nYour goal in this exercise is to build the best model, as a team, to predict the image's class in this dataset (more information in the Data tab). How you code and work as a team is up to you: you can all work on the same code and divide work, or build different models and ensemble amongs them, or any other method you prefer. Once you build the model, you need to upload your submissions. You will have access to labeled training data and unlabeled validation data. Your model will be evaluated based on the log-loss obtained on this validation data.
\n\nBased on your log-loss, you will be assigned a rank on the Kaggle Leaderboard. This way you are able to monitor your standing in this competition. Once the competition is closed, the person at the top of the leaderboard is the winner. But wait, there's a twist. We will hold out an additional subset of the data that your model will be evaluated on. You will be unable to see the scores on this set until the competition closes. This means rankings can change when the competition closes. Therefore, though you try to improve your rankings, remember to keep your model generalizable so that you can have the last laugh.
\n\nThere was a problem with how labels were specified in the data. They are now one-hot encoded. If you don't have access to these one-hot encoded data please re-download.
\n\nThe timeline has now been updated, the competition will run through midnight on Friday (4/17/2020) and your reports are due by midnight on Monday (4/20/2020)
", "isPublished": false, "hasIsPublished": false, "postTitle": "", "hasPostTitle": false, "mimeType": "", "hasMimeType": false }, { "name": "Evaluation", "content": "It is your job to predict the correct label of the traffic sign. For each Id in the test set, you must predict the label of the German traffic sign image.
\n\nSubmissions are evaluated on log-loss between the predicted value and the observed sales price
\n\n\nid\tlabels\n0\t40\n1\t3\n2\t19\n\n\n