Home
Welcome to the senior-project-spring-2022-AI-positioning-system wiki! Project components:
- AI position Plan
- Similarity of pictures
- Find text or numbers in the picture
- Check all city, state, and country name
- Test project & debug
Updated project plan:
Maybe need to do an OCR pipeline to clear the text.
Plan 1
Plan 2
Snapshot 2 Updated project plan: The code about finding the similarity of pictures is done. The image Identification & image search code is already tested. The Balance aHash, Image Binarization, and Compared aHash are finished but haven't been tested and debugged yet. The picture database is built. And the first group has 3 similar pictures of Yosemite's natural parky and one of the pictures has the print name on it.
List activities behind schedule: My finding text or numbers in the picture part is very hard because I try to solve the clear problem. But I still cannot fix it by OCR. Also, my similar part needs to finish the test.
Next: I will code the find the text part without a clear problem and use tesseract-OCR(web) to help me with this part.
Callenges: (find text & number): Still Learning about scanning the picture using Optical Character Recognition(OCR) problem. Try to use text detection, character segmentation, and Character classification to clear the text.
Special notes: There is currently a problem with the text in the scanner and the text &number extraction. My code cannot precise, and the location of the white block for the text. I research some Scan the picture using Optical Character Recognition (OCR) essay and paper to help me to solve the problem of clarity, but it does not work for now. The current progress about finding similarities is complete but not tested yet. The image database for the first group has been compiled, and there are three similar images. And the first image has print text.
Snapshot 3 Updated project plan: The code about finding the similarity of pictures is done. The code of find location is finished. I test it, and it can give the correct country and place location. For example, in the picture database, 1 will show the Yosemite national park and America.
List activities behind schedule: Finding text or numbers in the picture part is very hard. I change my mind and I use the tesseract-OCR to find the scan the message from the picture. I have already downloaded the web member and try to make my code work. But the text does not work. Only the number can find and print from pictures.
Next: I will code the main menu part and make sure all my code connects. And I will keep working with the OCR part.
Challenges: the tesseract-OCR web may not very fit my requirements. So, I need to keep changing my code to make sure my Optical Character Recognition(OCR) code can get clear text and number info from pictures.
Special notes: There is currently a problem with the text in the scanner and the text &number extraction. My find location part code is almost done and it can find the correct area location.
Snapshot 4 Updated project plan: create the main part and connect ORC code to the web. now the picture database can connect with the ORC code. updated new wiki, find text& numbers, and readme files. Moreover, I finish the test to find the location part by training model.
List activities behind schedule: Need to connect similar parts with the main code together and finish testing the find location part.
Next: I will keep connecting all the things. And I will keep working with the test OCR part.
Challenges: the tesseract-OCR web may not very fit my requirements. So, I need to keep changing my code to make sure my Optical Character Recognition(OCR) code can get clear text and number info from pictures.
Special notes: My find location part code is done and it can find the correct area location. And this location code is connected to the picture database, so it can find my system. But I still need to connect this with other parts to complete the main part.
Snapshot 5 Updated project: Change some detailed data(use the web to find lat&lon location ) for the find location part. Finish testing all the code and debug it. The OCR part is running well and it successfully connects with the similarity part. Moreover, I finish the poster presentation slides and output file for this project.
List activities behind schedule: All the part is complete.
Next: I will be doing my poster presentation on May 7.
Challenges: The training model is hard to connect with other parts. And it is possible the ORC cannot read the blurry messages.
Special notes: I post my training data and test data to find the location part to train the model and get the correct location. My output file includes all the detail for each part output picture.
