What is Wobot Hackathon?
Wobot, after achieving multiple milestones in the year 2020, has established itself as a fast-paced environment where growth and innovation are the driving factors. We take pride in the products we offer and the brilliantly gifted team members who help create those products.
After successfully cultivating two batches of our Product Engineers in the last two years, we are back with the Hackathon in the search of the third batch of talented and young professionals.
What does it offer?
Upon successful completion of the task and the Personal Interview with some of our most talented lot, the selected candidates will be offered a 6-month internship with Wobot in the capacity of a Computer Vision Intern. All good things must come to an end, however post the internship period, a glamourous PPO opportunity awaits you. So, maybe all good things don’t certainly end.
Dataset: Dataset Link
- Segment defect and provide the predicted segmentation masks on test set (.jpeg image).
- Segment defect using only samples without defect and the predicted segmentation masks on test set (.jpeg image).
In ‘.txt’ file each column represents the following:
[filename] [semi-major axis] [semi-minor axis] [rotation angle(The rotation angle is measured counterclockwise (positive angle))] [x-position of the centre of the ellipsoid] [y-position of the centre of the ellipsoid] (Note: x- and y-coordinates are given in MATLAB format, i.e. the origin is in the upper left corner of the image.)
Optional Hardware for training: https://colab.research.google.com/
- The dataset contains Training, Evaluation and Testing Samples. There are 120 total test samples, for which, the ground truth is not provided.
- Participants will submit the Jupyter Notebook along with the outputs on these 120 samples and save that in folder named ‘test_outputs/’ in ‘.jpeg’ format binary image (segmentation masks).
- Internal evaluation script will be used to compare the results on submitted results from the ‘test_outputs/’ folder and their respective ground truths. Evaluation metrics will include:
- TP, TN, FP, FN also known as confusion matrix.
- Intersection over Union (IOU) and Mask Intersection over Union (mIOU), which will give the score between 0-1.
- If IOU and mIOU score is greater than or equal to 0.5 then it will be considered detected.
- Jupyter Notebook(s) containing the pre-processing, training, evaluation and output generation code
- Zip file containing predicted segmentation masks in .jpeg format for the 120 test images
Sample: Confusion Matrix
|Predicted Positive||Predicted Negative|
In case of any doubts react out to us at email@example.com.
The last date to send your submission is 23rd December, Hurry up!