update: fix dockerfile for simple no-build stages, new: add logic for ocr pipeline and streamlit

Dockerfile

@@ -2,17 +2,17 @@ FROM python:3.9
 
 WORKDIR /code
 
+COPY . .
+
 RUN pip install --upgrade pip
-RUN pip install torch torchvision
+RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
 RUN pip install -U openmim
 RUN mim install mmengine
 RUN mim install mmcv
 RUN mim install mmdet
 
 RUN git clone https://github.com/open-mmlab/mmocr.git
-RUN cd mmocr
-RUN pip install -v -e .
-
-COPY . .
+RUN pip install -v -e /code/mmocr
 
-CMD ["python", "app.py"]
+CMD ["streamlit", "run", "app.py"]

app.py

@@ -1,5 +1,17 @@
+from datasets import load_dataset
+import streamlit as st
 from mmocr.apis import MMOCRInferencer
+import mmcv
+import matplotlib.pyplot as plt
 
-infer = MMOCRInferencer(rec='svtr-small')
-result = infer('demo/demo_text_recog.jpg', save_vis=True, return_vis=True)
-print(result['predictions'])
+st.write('Hello, *World!* :sunglasses:')
+dataset = load_dataset("katanaml-org/invoices-donut-data-v1")
+st.write(f"{len(dataset)}")
+
+infer = MMOCRInferencer(det='dbnetpp', rec='svtr-small', kie='SDMGR')
+result = infer('demo/demo_kie.jpeg', save_vis=True)
+
+predicted_img = mmcv.imread('results/vis/demo_kie.jpg')
+plt.figure(figsize=(18, 32))
+plt.imshow(mmcv.bgr2rgb(predicted_img))
+plt.show()

requirements.txt

@@ -0,0 +1,5 @@
+torch
+torchvision
+datasets
+streamlit
+matplotlib