One of the biggest challenges to perfecting self-driving cars is navigating varying weather conditions. Much like our eyes, car sensors don't work as well in fog, rain or snow.
For this puzzle, your model will have to classify images from moving self-driving cars in one of the five categories. The categories vary from a clear sky in daylight to heavy rails and wet roads in dark night sky.
The five weather categories are as follows:
1) Weather is clear
2) Road is slightest wet & weather little bit cloudy
3) Wet road, cloudy weather, and light rains
4) Rainy weather at dusk or dawn time.
5) Heavy rains & wet roads at night.
Create a model that classifies input images in one of the five categories. Check out the starter kit to know the first steps.
✔ The Task
The challenge is to use the images from the dataset to build an automated algorithm to classify images into 5 different classes as mentioned above without any labels:
In machine learning terms: this is unsupervised image classification.
🚀 Getting Started
Make your first submission using starter kit. 🚀
The dataset contains images from different weather settings generated using the Carla Simulator and your task will be to classify the weather into 5 different categories without any labels.
Following files are available in the
data.zip- ( 700 samples ) - The zip file contains the models that you will classify into 5 categories.
📨 How to submit
Make your first submission using the starter kit 🚀
Save the submission.csv in the assets directory. The name of the above file should be
Inside a submission directory, put the
.ipynbnotebook from which you trained the model and made inference and save it as
Zip the submission directory
Overall, this is what your submission directory should look like -
submission.zip ├── assets │ └── submission.csv └── original_notebook.ipynb
Make your first submission here,
🖊 Evaluation Criteria
The evaluation metric for this competition is the Adjusted Rand Score.
Resnet50 + Kmeans based image clustering model
[Getting Started Notebook] Environment Classification