Skip to main content
added 148 characters in body
Source Link
Becky 李蓓
  • 17.1k
  • 13
  • 53
  • 170

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

We can open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs; fix bugs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

(It may be useful here to search by radical to identify troublesome characters.)

###Step 4: Translate.

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

We can open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs; fix bugs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

###Step 4: Translate.

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

We can open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs; fix bugs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

(It may be useful here to search by radical to identify troublesome characters.)

###Step 4: Translate.

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

added 40 characters in body
Source Link
Becky 李蓓
  • 17.1k
  • 13
  • 53
  • 170

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

IWe can open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputsoutputs; fix bugs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

###Step 4: Translate.

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

I open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

We can open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs; fix bugs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

###Step 4: Translate.

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

added 77 characters in body
Source Link
Becky 李蓓
  • 17.1k
  • 13
  • 53
  • 170

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

I open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

IfOCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs.

We can compare these two transcriptions. It's The writing is stuck on athe inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

I open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs.

We can compare these two transcriptions. It's stuck on a inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

##Handwriting

TL;DR: Handwriting is much harder than printed text. The first step is to improve the image quality. Afterwards we can input the text into OCR software.

This is on a door where I'm staying.

If we put it directly into Online OCR, it says:

No recognized text !

The image quality is too poor.

###Step 1: Improve the image quality.

I open up Paint (actually I use Pinta) and go over the Chinese in black.

Method 1: change the brightness and contrast:

Method 2: select only the black text (e.g. GIMP has Tools/Selection Tools/By Color Select):

###Step 2: Input into OCR software.

OCR software struggles with handwriting, so we may need to try a few times. If we input the second image above into Online OCR, we obtain:

请各位出入
头好蓝色水门

It made two mistakes:

  • the original says 关好 (close) and not 头好 (meaningless);
  • the original says 木门 (timber door) and not 水门 (water gate).

So it's not perfect.

However, inputting the third image into EasyScreenOCR gives:

请各位出入
关好益包木门

It also made two errors:

  • the original says 蓝色 (blue) and not 益包 (meaningless).

###Step 3: Compare the outputs.

We can compare these two transcriptions. The writing is stuck on the inside of a blue wooden door, so...

  • (to close) makes sense and not 头 (head);
  • 蓝色 (blue) makes sense;
  • 木门 (timber door) makes sense.

So we can fix the bugs to obtain:

请各位出入
关好蓝色木门

Now we put this into Google Translate to give:

Please come in and close the blue wooden door

Actually, it's not a very good translation (出入 means "come in and go out"). We can break this up a bit further using Yabla:

Thus it realistically means something like:

Please, when coming in or going out, close the blue wooden door.

Source Link
Becky 李蓓
  • 17.1k
  • 13
  • 53
  • 170
Loading