Detect Duplicate Image Series by
Budget: $10 – $40 USD
Task: Create a demo code solution for the following: Detect Duplicate Image Series by location
Time Limit: 10 hours.
Type of Task: Back-End
How the Code Should Be Made:
- It should meet these requirements: We have a series of images taken along roadways at regular distance intervals. After we take all of the images, we want to identify any places in the series of images where we drove over a road we already captured, so we can separate and archive them.
Input: Numpy array with an image name, latitude, longitude, and date taken, like so:
[('Image_000001.jpg', 40.87066159, 101.72716008, '2023:10:30 15:01:25')
('Image_000002.jpg', 40.8706409 , 101.72704223, '2023:10:30 15:01:27')
('Image_000003.jpg', 40.87063224, 101.72692113, '2023:10:30 15:01:29')
...
('Image_003805.jpg', 40.84026789, 101.72362902, '2023:10:30 17:29:53')
('Image_003806.jpg', 40.84017844, 101.72362409, '2023:10:30 17:29:56')
('Image_003807.jpg', 40.84009071, 101.72365949, '2023:10:30 17:30:04')]
Explanation of method:
1. Identify the average distance between images in the data set, using the latitude and longitude to calculate. Then, calculate the standard deviation of these distances.
AverageDistance = Avg(Distance[])
StdDevDistance = StdDev(Distance[])
2. With that average distance between images, identify any points which are which are closer than the AverageDistance - 1*StdDevDistance to another point. These points are “candidate duplicates”.
3. The candidate duplicates should occur in consecutive runs (ie 6 consecutive images at 11:56AM are duplicates with 6 consecutive images at 12:45PM). For each point in the candidate duplicates list, identify which “run” it is a part of and identify the later run (the overlap that occurred later in time) as “duplicates”
Return the list of duplicate file names.
- Use these tools/languages: Python 3.12
End Result:
- Functional Code Solution
- Steps on how to use the code, written clearly
Please add lots of comments inside of the code explaining your work.
If your solution requires any third-party services that need accounts, access keys, or API keys, please create your own accounts using free or trial versions. Clearly mention which keys need to be replaced in your README file.
Time Limit: 10 hours.
Type of Task: Back-End
How the Code Should Be Made:
- It should meet these requirements: We have a series of images taken along roadways at regular distance intervals. After we take all of the images, we want to identify any places in the series of images where we drove over a road we already captured, so we can separate and archive them.
Input: Numpy array with an image name, latitude, longitude, and date taken, like so:
[('Image_000001.jpg', 40.87066159, 101.72716008, '2023:10:30 15:01:25')
('Image_000002.jpg', 40.8706409 , 101.72704223, '2023:10:30 15:01:27')
('Image_000003.jpg', 40.87063224, 101.72692113, '2023:10:30 15:01:29')
...
('Image_003805.jpg', 40.84026789, 101.72362902, '2023:10:30 17:29:53')
('Image_003806.jpg', 40.84017844, 101.72362409, '2023:10:30 17:29:56')
('Image_003807.jpg', 40.84009071, 101.72365949, '2023:10:30 17:30:04')]
Explanation of method:
1. Identify the average distance between images in the data set, using the latitude and longitude to calculate. Then, calculate the standard deviation of these distances.
AverageDistance = Avg(Distance[])
StdDevDistance = StdDev(Distance[])
2. With that average distance between images, identify any points which are which are closer than the AverageDistance - 1*StdDevDistance to another point. These points are “candidate duplicates”.
3. The candidate duplicates should occur in consecutive runs (ie 6 consecutive images at 11:56AM are duplicates with 6 consecutive images at 12:45PM). For each point in the candidate duplicates list, identify which “run” it is a part of and identify the later run (the overlap that occurred later in time) as “duplicates”
Return the list of duplicate file names.
- Use these tools/languages: Python 3.12
End Result:
- Functional Code Solution
- Steps on how to use the code, written clearly
Please add lots of comments inside of the code explaining your work.
If your solution requires any third-party services that need accounts, access keys, or API keys, please create your own accounts using free or trial versions. Clearly mention which keys need to be replaced in your README file.