Free! Batch! No need to connect to the internet! Extract text from images with this app!

Which method do you use to extract text from images? WeChat, some paid OCR software, or typing it manually? Here, I would like to introduce a free software that can batch extract text from images without the need for an internet connection - Umi-OCR.

Umi-OCR is relatively easy to use. You can download it and unzip it to use. By default, it only supports simplified Chinese and English. After downloading the expansion pack, it can support recognition operations of Traditional Chinese, Japanese, Korean, German, and French.

​​​​​

Figure 1 You can download the expansion pack to support more language recognition

​​Umi-OCR Download Address​​

Simple interface, try to extract text from pictures

Open the Umi-OCR main interface, you can see a simple and easy-to-use operation interface. Users only need to click the "Browse" button or directly drag the picture that needs to be recognized text to the Umi-OCR window to import the picture, and then press the "Start Task" button to extract text from the picture. It supports batch image recognition and does not require an Internet connection.

​​​​​

Figure 2 Umi-OCR main interface

The recognition accuracy of Umi-OCR is quite good. For example, in the unclear newspaper image below, Umi-OCR can correctly recognize and extract most of the text.

​​​​​

Figure 3 Example

​​​​​

Figure 4 Umi-OCR recognition results

For clear images like the one below, Umi-OCR has a higher recognition rate. Basically, the extracted and recognized text content can be used without modification.

​​​​​

Figure 5 Example

​​​​​

Figure 6 Umi-OCR recognition results

No need to save locally Clipboard recognition function

Umi-OCR supports clipboard reading function. When you see a picture while browsing the web, you don't need to save it before performing the recognition operation. Just copy the picture and click clipboard reading in Umi-OCR, and Umi-OCR will complete the recognition operation. Of course, you can also use this function to perform text recognition directly after taking a screenshot.

Ignore watermarks Identify area management functions

If there is a watermark on the image, the text content with the watermark will be recognized, and you have to delete it one by one, which is too troublesome. Umi-OCR has prepared an ignore watermark function for users, which can automatically ignore the text recognition in the specified area when recognizing text.

To enable the ignore watermark function, you must first import the image to be recognized, and then select "Add Area" in the Umi-OCR setting options. At this time, an area setting will appear. At this time, you only need to select the watermark area to ignore the text recognition in that area.

​​​​​

Figure 7 Adding a region

​​​​​

Figure 8 Identification area settings

Summarize

Umi-OCR uses the PPOCR-v3 model library by default. In addition to being able to accurately recognize regular text, it also has a good recognition rate for unconventional fonts (handwriting, artistic characters, small characters, irregular directions, messy backgrounds, etc.). It does not require an Internet connection to upload images for recognition, and its performance is quite good. Netizens who need it can try it.