OCR with AI (intelligend data processing)
Budget: $500 – $1,000 USD
I already have a python software source-code that converts PDF files into structured Json files.
This is done using OCR and YOLOv5.
The PDF files are supplier invoices. These invoices consist of a header, x items and a footer. There are a subset of predfinded variables (like customer_number, invoice_number, ...) to extract from pdf.
I now need the following adjustments:
1.) the tool needs to be improved. Because it does not recognize the data correctly for the positions.
2.) I need a simple graphical user interface to learn (train) more PDF files. I need a generic solution for this. Today's solution is not generic, but trained and coded for 3 different PDF formats.
3.) the learned models should learn from each other. This means that the tool should learn automatically when new PDF files are trained. If a new format comes along, it should also recognize it as best as possible without necessarily having to train it.
Now I am asking for offers to improve the tool and bring in these 3 points.
These are the requirements of the python tool that must be installed for the tool to run:
# YOLOv5 requirements
# Usage: pip install -r requirements.txt
# bases ----------------------------------------
flask
easyocr
pdf2image
image slicer
matplotlib>=3.2.2
numpy>=1.18.5
opencv-python>=4.1.1
Pillow>=7.1.2
PyYAML>=5.3.1
requests>=2.23.0
scipy>=1.4.1
torch>=1.7.0 # see https://pytorch.org/get-started/locally/ (recommended)
torchvision>=0.8.1
tqdm>=4.64.0
# protobuf<=3.20.1 # https://github.com/ultralytics/yolov5/issues/8012
# logging -------------------------------------
tensorboard>=2.4.1
# wallb
# clearml
# Plot ------------------------------------
pandas>=1.1.4
seaborn>=0.11.0
# Export -------------------------------------
# coremltools>=6.0 # CoreML export
# onnx>=1.9.0 # ONNX export
# onnx-simplifier>=0.4.1 # ONNX simplifier
# nvidia-pyindex # TensorRT export
# nvidia-tensorrt # TensorRT export
# scikit-learn<=1.1.2 # CoreML quantization
# tensorflow>=2.4.1 # TF exports (-cpu, -aarch64, -macos)
# tensorflowjs>=3.9.0 # TF.js export
# openvino-dev # OpenVINO export
# Deploy --------------------------------------
# tritonclient[all]~=2.24.0
# Tools --------------------------------------
ipython # interactive notebook
psutil # system utilization
thop>=0.1.1 # FLOP's computation
# mss # screenshots
# album entries>=1.0.3
# pycocotools>=2.0 # COCO mAP
# roboflow
This is done using OCR and YOLOv5.
The PDF files are supplier invoices. These invoices consist of a header, x items and a footer. There are a subset of predfinded variables (like customer_number, invoice_number, ...) to extract from pdf.
I now need the following adjustments:
1.) the tool needs to be improved. Because it does not recognize the data correctly for the positions.
2.) I need a simple graphical user interface to learn (train) more PDF files. I need a generic solution for this. Today's solution is not generic, but trained and coded for 3 different PDF formats.
3.) the learned models should learn from each other. This means that the tool should learn automatically when new PDF files are trained. If a new format comes along, it should also recognize it as best as possible without necessarily having to train it.
Now I am asking for offers to improve the tool and bring in these 3 points.
These are the requirements of the python tool that must be installed for the tool to run:
# YOLOv5 requirements
# Usage: pip install -r requirements.txt
# bases ----------------------------------------
flask
easyocr
pdf2image
image slicer
matplotlib>=3.2.2
numpy>=1.18.5
opencv-python>=4.1.1
Pillow>=7.1.2
PyYAML>=5.3.1
requests>=2.23.0
scipy>=1.4.1
torch>=1.7.0 # see https://pytorch.org/get-started/locally/ (recommended)
torchvision>=0.8.1
tqdm>=4.64.0
# protobuf<=3.20.1 # https://github.com/ultralytics/yolov5/issues/8012
# logging -------------------------------------
tensorboard>=2.4.1
# wallb
# clearml
# Plot ------------------------------------
pandas>=1.1.4
seaborn>=0.11.0
# Export -------------------------------------
# coremltools>=6.0 # CoreML export
# onnx>=1.9.0 # ONNX export
# onnx-simplifier>=0.4.1 # ONNX simplifier
# nvidia-pyindex # TensorRT export
# nvidia-tensorrt # TensorRT export
# scikit-learn<=1.1.2 # CoreML quantization
# tensorflow>=2.4.1 # TF exports (-cpu, -aarch64, -macos)
# tensorflowjs>=3.9.0 # TF.js export
# openvino-dev # OpenVINO export
# Deploy --------------------------------------
# tritonclient[all]~=2.24.0
# Tools --------------------------------------
ipython # interactive notebook
psutil # system utilization
thop>=0.1.1 # FLOP's computation
# mss # screenshots
# album entries>=1.0.3
# pycocotools>=2.0 # COCO mAP
# roboflow