The objective of this hackathon is to extract highly unstructured data and use that information to generate insights!
A simple example is counting the number of cars in parking lots for a company and in turn forecasting sales for that company.
Can you use the datasets provided below to “nowcast” economic trends?
Apart from the awesome cash prizes you could win, you could also get hired with Granular Data.
Get hired with granular.ai for a whooping INR 16L-18L per annum and an incredible share in equity!
The role would require you to work remotely.
Click here to check out the granular.ai team and detailed Job Description.
This competition is open to people of all skill levels, but only people with prior knowledge of deep learning will be considered for the full time position. That said, we will reward creative ideas :)
Earth Observation Satellite Data
Spacenet Dataset
Sentinel Data
Landsat Dataset
USGS EarthExplorer
Planet Labs (California, 4 band, medium resolution)
GBDX by digital globe (S.Africa, San Fransisco, 3 and 8 band, high resolution)
Planet Labs and GBDX have restrictions on where you can use the data, so downloading for your own model may present extra hurdles!
Other data sets that could be useful
Geo - Metadata - incredibly valuable for data labels
- Open Street Maps, OSM humanitarian
- Google Maps
Financial Data/Economic Data
While we definitely do not want to discourage any participants, deep learning is a pretty tough topic to pick up in a few weeks time :( That said, if you have experience with neural networks and are looking to better understand how to work with computer vision problems and satellite images, or the specific models that work best, REGISTER NOW to unlock the Resource Center filled with Tutorials, Articles, Videos and Papers!
What tools can you use for building your solution?
This hackathon is completely language/library agnostic - use whatever you feel most comfortable with :)
What do you need to submit?
Submit all code and any presentation you have established on the importance of that code!
Presentation does not have to be a formal slide deck, could be a Jupyter Notebook with stunning visuals or a link to a dashboard. You can even submit an API!
This is what you'll be evaluated on:
- Idea
- Implementation
- Presentation
How will we evaluate submissions?
Your submission should prove that you can build and develop deep learning models.
If your idea is really ambitious and won’t show this on its own, maintain a side project that object detection or semantic segmentation.
Idea creativity will also go a long way. Granular is a small team and contending against big companies, so out of the box thinking is a prerequisite to join the team. If you can help generate the next product idea, Granular will work with you or buy your concept outright, even if you do not win!
Satellite Imagery
Orbital Insights: Retail Car Counting, Oil tanks
Spaceknow: Industrial production
We really like this concept and would love to see similar ideas :)
Economic Empowerment using unconventional data
Premise (awesome company!)
Tala
Nowcast (monitoring GDP with night light data and store transactions)
Click here to fill up the form and get access to AWS Credits!
