I want a program written in Python that takes a screenshot of a webpage, and can return the location of banner and video ads on a page as a list of (x,y) coordinates or one singular coordinate.
would return: [(916, 392), (927, 988)]
Where the first tuple is the 728x90 banner at the top of the page, and the second tuple is the longer, skinny bar at the bottom of the page.
The second image (img-2) would return: (1574, 818) to ID the 300x250 banner in the lower-left area of the screen.
The third image (img-3) would return three tuples in a list: [(1136, 763), (353, 698), (933, 288)]
--one for the video ad in the center, and two for the banners.
If no banner is found, then it should return a string saying "no banner found."
The program should work on all websites, not just the WSJ. So there will have to be some level of machine learning component to train a model to identify all of the banners / videos. This should also take into account banners that are not fully on the screen--meaning the images won't always be the perfect ad sizes.
The task will require you to use some combination of sklearn, keras, tensorflow, numpy, opencv and whatever other libraries you think would work best.
The final deliverable will be a working program that I can deploy in a production environment--likely set up in a github repo. I don't pay for programs that I haven't run and extensively bug tested, and I expect you to test it before sending it into me.
I have a lot of work along this front, so if you do a good job, there will be more jobs after this.
I look forward to working with you.
22 freelancer đang chào giá trung bình $518 cho công việc này
Hi there, I have checked the details I have great experience with Keras, Machine Learning, OpenCV, Python, Tensorflow. Please start the chat so we can discuss this job more in detail. Thanks