Inspiration
This is a project built to complete a challenge named - "Game of Drones". The main idea of the challenge is to encourage projects which can be helpful in identifying different bodies and objects on the surface of moon using well-equipped drone. We found the challenge quite interesting and intriguing as it will involve Machine Learning and Image Processing concepts to achieve the project.
What it does
An arena is provided by the organizers. The drone flies above the arena, captures pictures/videos and identifies the different objects.
How we built it
The drone was given by the organizers which was equipped with a high-definition camera. We created a data set of all the objects in the arena. We, then used feature extraction concepts in OpenCv and processed the images in the data set. Later, we trained the data set using Tensorflow model to be able to classify the data-set with appropriate labels.
Challenges we ran into
- Our environment was being crashed due to high-level processes.
- It is quite a task to detect objects in a video feed.
- Considering the surrounding disturbances (like dim light, unwanted objects) in the recorded photos/video, it was difficult to be able to detect the required objects. w
Accomplishments that we're proud of
- We could operate the drone smoothly and efficiently.
- We could classify two types of objects successfully.
- In-spite of many difficulties, we did not give up on the project and tried to give our best.
What we learned
- We learned the basic operation of drone.
- We analysed different object detection algorithms.
- We learnt the Machine Learning tool, Tensorflow.
What's next for Game of Drones
- To achieve the detection of objects in a real-time i.e, detecting the objects in the real-time video captured while flying the drone.
Log in or sign up for Devpost to join the conversation.