Text detection with Python | Tesseract vs Easyocr vs AWS Textract | What is the best OCR?

[@ComputerVisionEngineer]

Did you enjoy this video? Try my premium courses! 😃🙌😊 ● Hands-On Computer Vision in the Cloud: Building an AWS-based Real Time Number Plate Recognition System bit.ly/3RXrE1Y ● End-To-End Computer Vision: Build and Deploy a Video Summarization API bit.ly/3tyQX0M ● Computer Vision on Edge: Real Time Number Plate Recognition on an Edge Device bit.ly/4dYodA7 ● Machine Learning Entrepreneur: How to start your entrepreneurial journey as a freelancer and content creator bit.ly/4bFLeaC Learn to create AI-based prototypes in the Computer Vision School! www.computervision.school 😃🚀🎓

[@hugoleyva1741]

I'll be waiting for that video tysm 😌

[@Intellectualmind4]

Keep it up 🎉🎉🎉🎉

[@ComputerVisionEngineer]

😃🙌

[@bauyrzhan8652]

Hello. Thank you for your tutorials. What is best OCR module for the licence plates recognition for all types of plates (not like your previous video Yolov8, where you had function license_complies_format )? The car stops before a barrier opens and so that the plates is recorded at rest.

[@ComputerVisionEngineer]

Easyocr and AWS textract are good options. You may need to pre process the image in order to improve your results. 🙌

[@Matheus-kk9qh]

Hello sir, thanks for another video, what do you think about mmocr? And how do you keep informed about new technologies, models, architectures and so on, in Computer Vision field ?

[@summa7545]

Hi, first of all I'd like to appreciate your efforts. Your works help me learn CV easily. Is it possible to do a video on handwritten text detection? We have videos on text detection on licence plates, documents etc. But handwritten detection is not done anywhere. So far all tools like paddleocr, easyocr, tesseract etc fails to capture all data correctly for handwritten text. It would be very helpful if you do a video on this. 😊

[@greyswandir2807]

Can you make a tutorial to show us how to select an area to scan (as opposed to the whole screen/image)?

[@McMedLife]

I'll try to explain the best I can within the confines of KZbin comments section. Roughly speaking: - Read your image with OpenCV (imread). - Files read via OpenCV are numpy arrays, so you can easily select the "area" you want. - E.g. say you read an image of width 1000, height 700 and you want to only process the very middle of the photo, with a width and height of 100. - Calculating the coordinates of your desired crop: x1 = (1000-100)/2 which is 450. This will be the start of your x (450) for your desired width (100) which makes the end point 450+100=550 y1 = (700-100)/2 which is 300. The same concept here. Start of your y will be 300 and the end at 400. - Keep in mind that numpy arrays are indexed with your rows (ie "heigh") first, then columns (ie "width") and finally your channels. So, assuming your image is width 1000 height 700 and a colour image (eg RGB), it would have a shape of (700,1000,3)! - So after you read your image with OpenCV (eg image = cv2.imread(file_path)), do any processing you have to do, and then create a subset of your image like: image_cropped = image[300:400, 450:550] **** remember you're putting the rows (ie height) first!! That's it. image_cropped is now the crop you wanted. It is a cropped OpenCV version of your image which is essentially a numpy array. Now, depending on where you're trying to feed it to and how you're planning to submit it for text detection, you can save it (eg with imwrite) or convert it to other types. For example, you may want to convert it to a Pillow Image type (which, after doing "from PIL import Image", you can do with Image.fromarray(image_cropped). One more thing to keep in mind when working with OpenCV, is the channels and their "format" such as RGB vs. BGR vs. others. Hope this helps!

[@atahirince]

could be nice to see response time performance

[@litziadrianacruz7583]

Hi! Thanks for the video. What do you think of MaskOCR?

[@ComputerVisionEngineer]

Haven't had the time to get involved with MaskOCR, I will try to take a look at it. 🙌

[@rohanshah8129]

Thanks for this informative video!!! How about video 2 video project? Convert an video into Anime style video???? :D

[@ComputerVisionEngineer]

Video style transfer sounds like a cool idea for a future video! I will try to do it. 🙌

[@undertaker199411]

Thanks for the vid.. 🙂 But you should not show your AWS access_key and secret_access_key in an KZbin Video, it may be dangerous

[@ComputerVisionEngineer]

Glad you enjoyed it! No problem, I deleted the access keys after making the video. 🙌

[@arfanislamabir]

data set are not available on petron. and there is lot person like who doesn't have credit card or ability to pay you. you are making great video but limited for people. I am leaving your channel

[@ComputerVisionEngineer]

I appreciate your feedback. Creating content involves significant time and effort, and sharing additional content with my Patreons helps sustain this creative endeavor. Regarding the dataset you are referring to, it is also possible to create it yourself by downloading images from pexels and then manually annotating them. 🙌