A repository for hosting all of the OCR code written for Aadhar, PAN and GSTIN
Unzip the inference_graph.zip file in the inference_graph_aadhar directory.
python Object_detection_image.py --image <image_path>
python Object_detection_webcam.py
Unzip the ckpt.zip file in the same directory.
python YoloDemo_image.py --image <image_path>
python YoloDemo_webcam.py
Use generate_tfrecord.py to generate tfrecord files for test and training set from labeled images.
Use train.py to train the tensorflow model for any change in format.
For labeling images follow the *github link.
For detailed explanation you can follow first reference.
The training for yolo goes similarly except for the model you choose.
The tensorflow model is trained on ssd_mobilenet_v1 model which is computationally light with pretty accurate results.
- Tensorflow Object Detection API - tutorial
- YOLO Detect - tutorial