Description
For table detection it is proposed the use of CascadeTabNet. It is a Cascade mask Region-based CNN High-Resolution Network (Cascade mask R-CNN HRNet) based model that detects tables on input images. The model is evaluated on ICDAR 2013, ICDAR 2019 and TableBank public datasets. It achieved 3rd rank in ICDAR 2019 post-competition results for table detection while attaining the best accuracy results for the ICDAR 2013 and TableBank dataset.
Here it is used the CascadeTabNet general model for table detection inspired by https://arxiv.org/abs/2004.12629
Predicted Entities
Live Demo Open in Colab Copy S3 URI
How to use
binary_to_image = BinaryToImage() \
.setInputCol("content") \
.setOutputCol("image") \
.setImageType(ImageType.TYPE_3BYTE_BGR)
table_detector = ImageTableDetector.pretrained("general_model_table_detection_v2", "en", "clinical/ocr") \
.setInputCol("image") \
.setOutputCol("table_regions")
draw_regions = ImageDrawRegions() \
.setInputCol("image") \
.setInputRegionsCol("table_regions") \
.setOutputCol("image_with_regions") \
.setRectColor(Color.red)
pipeline = PipelineModel(stages=[
binary_to_image,
table_detector,
draw_regions
])
# Download image:
# !wget -q https://github.com/JohnSnowLabs/spark-ocr-workshop/raw/4.0.0-release-candidate/jupyter/data/tab_images/cTDaR_t10168.jpg
imagePath = "cTDaR_t10168.jpg"
image_df = spark.read.format("binaryFile").load(imagePath)
result = pipeline.transform(image_df)
val binary_to_image = new BinaryToImage()
.setInputCol("content")
.setOutputCol("image")
.setImageType(ImageType.TYPE_3BYTE_BGR)
val table_detector = ImageTableDetector.pretrained("general_model_table_detection_v2", "en", "clinical/ocr")
.setInputCol("image")
.setOutputCol("table_regions")
val draw_regions = new ImageDrawRegions()
.setInputCol("image")
.setInputRegionsCol("table_regions")
.setOutputCol("image_with_regions")
.setRectColor(Color.red)
val pipeline = new PipelineModel().setStages(Array(
binary_to_image,
table_detector,
draw_regions))
# Download image:
# !wget -q https://github.com/JohnSnowLabs/spark-ocr-workshop/raw/4.0.0-release-candidate/jupyter/data/tab_images/cTDaR_t10168.jpg
val imagePath = "cTDaR_t10168.jpg"
val image_df = spark.read.format("binaryFile").load(imagePath)
val result = pipeline.transform(image_df)
Example
Model Information
| Model Name: | ocr_table_detection_general_model |
| Type: | ocr |
| Compatibility: | Visual NLP 3.0.0+ |
| License: | Licensed |
| Edition: | Official |
| Output Labels: | [table regions] |
| Language: | en |
Benchmarking
3rd rank in ICDAR 2019 post-competition
1st rank in ICDAR 2013 and TableBank dataset