unir/MastersThesis
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Paddle ocr gpu support. #4

Merged
Seryusjj merged 40 commits from gpu_support into main 2026-01-19 17:35:25 +00:00
Conversation 0 Commits 40 Files Changed 105 +754 -1290
19 changed files with 754 additions and 1290 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 68efb27a1e - Show all commits

2
.gitignore vendored
View File

@@ -8,3 +8,5 @@ results
node_modules
src/paddle_ocr/wheels
src/*.log
src/output_*.ipynb
debugset/

29
src/dataset_manager.py
View File

@@ -43,3 +43,32 @@ class ImageTextDataset:
text = f.read()
return image, text
def get_output_path(self, idx, output_subdir, debugset_root="/app/debugset"):
"""Get output path for saving OCR result to debugset folder.
Args:
idx: Sample index
output_subdir: Subdirectory name (e.g., 'paddle_text', 'doctr_text')
debugset_root: Root folder for debug output (default: /app/debugset)
Returns:
Path like /app/debugset/doc1/{output_subdir}/page_001.txt
"""
img_path, _ = self.samples[idx]
# img_path: /app/dataset/doc1/img/page_001.png
# Extract relative path: doc1/img/page_001.png
parts = img_path.split("/dataset/", 1)
if len(parts) == 2:
rel_path = parts[1] # doc1/img/page_001.png
else:
rel_path = os.path.basename(img_path)
# Replace /img/ with /{output_subdir}/
rel_parts = rel_path.rsplit("/img/", 1)
doc_folder = rel_parts[0] # doc1
fname = os.path.splitext(rel_parts[1])[0] + ".txt" # page_001.txt
out_dir = os.path.join(debugset_root, doc_folder, output_subdir)
os.makedirs(out_dir, exist_ok=True)
return os.path.join(out_dir, fname)

111
src/doctr_raytune_rest.ipynb Normal file
View File

@@ -0,0 +1,111 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "header",
"metadata": {},
"source": [
"# DocTR Hyperparameter Optimization via REST API\n",
"\n",
"Uses Ray Tune + Optuna to find optimal DocTR parameters.\n",
"\n",
"## Prerequisites\n",
"\n",
"```bash\n",
"cd src/doctr_service\n",
"docker compose up ocr-cpu # or ocr-gpu\n",
"```\n",
"\n",
"Service runs on port 8003."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "deps",
"metadata": {},
"outputs": [],
"source": [
"%pip install -q -U \"ray[tune]\" optuna requests pandas"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "setup",
"metadata": {},
"outputs": [],
"source": [
"from raytune_ocr import (\n",
" check_workers, create_trainable, run_tuner, analyze_results, correlation_analysis,\n",
" doctr_payload, DOCTR_SEARCH_SPACE, DOCTR_CONFIG_KEYS,\n",
")\n",
"\n",
"# Worker ports\n",
"PORTS = [8003]\n",
"\n",
"# Check workers are running\n",
"healthy = check_workers(PORTS, \"DocTR\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "tune",
"metadata": {},
"outputs": [],
"source": [
"# Create trainable and run tuning\n",
"trainable = create_trainable(PORTS, doctr_payload)\n",
"\n",
"results = run_tuner(\n",
" trainable=trainable,\n",
" search_space=DOCTR_SEARCH_SPACE,\n",
" num_samples=64,\n",
" num_workers=len(healthy),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "analysis",
"metadata": {},
"outputs": [],
"source": [
"# Analyze results\n",
"df = analyze_results(\n",
" results,\n",
" prefix=\"raytune_doctr\",\n",
" config_keys=DOCTR_CONFIG_KEYS,\n",
")\n",
"\n",
"df.describe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "correlation",
"metadata": {},
"outputs": [],
"source": [
"# Correlation analysis\n",
"correlation_analysis(df, DOCTR_CONFIG_KEYS)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"name": "python",
"version": "3.10.0"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

29
src/doctr_service/dataset_manager.py
View File

@@ -43,3 +43,32 @@ class ImageTextDataset:
text = f.read()
return image, text
def get_output_path(self, idx, output_subdir, debugset_root="/app/debugset"):
"""Get output path for saving OCR result to debugset folder.
Args:
idx: Sample index
output_subdir: Subdirectory name (e.g., 'paddle_text', 'doctr_text')
debugset_root: Root folder for debug output (default: /app/debugset)
Returns:
Path like /app/debugset/doc1/{output_subdir}/page_001.txt
"""
img_path, _ = self.samples[idx]
# img_path: /app/dataset/doc1/img/page_001.png
# Extract relative path: doc1/img/page_001.png
parts = img_path.split("/dataset/", 1)
if len(parts) == 2:
rel_path = parts[1] # doc1/img/page_001.png
else:
rel_path = os.path.basename(img_path)
# Replace /img/ with /{output_subdir}/
rel_parts = rel_path.rsplit("/img/", 1)
doc_folder = rel_parts[0] # doc1
fname = os.path.splitext(rel_parts[1])[0] + ".txt" # page_001.txt
out_dir = os.path.join(debugset_root, doc_folder, output_subdir)
os.makedirs(out_dir, exist_ok=True)
return os.path.join(out_dir, fname)

2
src/doctr_service/docker-compose.yml
View File

@@ -14,6 +14,7 @@ services:
- "8003:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- doctr-cache:/root/.cache/doctr
environment:
- PYTHONUNBUFFERED=1
@@ -35,6 +36,7 @@ services:
- "8003:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- doctr-cache:/root/.cache/doctr
environment:
- PYTHONUNBUFFERED=1

7
src/doctr_service/doctr_tuning_rest.py
View File

@@ -169,6 +169,7 @@ class EvaluateRequest(BaseModel):
# Page range
start_page: int = Field(5, ge=0, description="Start page index (inclusive)")
end_page: int = Field(10, ge=1, description="End page index (exclusive)")
save_output: bool = Field(False, description="Save OCR predictions to debugset folder")
class EvaluateResponse(BaseModel):
@@ -302,6 +303,12 @@ def evaluate(request: EvaluateRequest):
)
time_per_page_list.append(float(time.time() - tp0))
# Save prediction to debugset if requested
if request.save_output:
out_path = state.dataset.get_output_path(idx, "doctr_text")
with open(out_path, "w", encoding="utf-8") as f:
f.write(pred)
m = evaluate_text(ref, pred)
cer_list.append(m["CER"])
wer_list.append(m["WER"])

111
src/easyocr_raytune_rest.ipynb Normal file
View File

@@ -0,0 +1,111 @@
{
"cells": [
{
"cell_type": "markdown",
"id": "header",
"metadata": {},
"source": [
"# EasyOCR Hyperparameter Optimization via REST API\n",
"\n",
"Uses Ray Tune + Optuna to find optimal EasyOCR parameters.\n",
"\n",
"## Prerequisites\n",
"\n",
"```bash\n",
"cd src/easyocr_service\n",
"docker compose up ocr-cpu # or ocr-gpu\n",
"```\n",
"\n",
"Service runs on port 8002."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "deps",
"metadata": {},
"outputs": [],
"source": [
"%pip install -q -U \"ray[tune]\" optuna requests pandas"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "setup",
"metadata": {},
"outputs": [],
"source": [
"from raytune_ocr import (\n",
" check_workers, create_trainable, run_tuner, analyze_results, correlation_analysis,\n",
" easyocr_payload, EASYOCR_SEARCH_SPACE, EASYOCR_CONFIG_KEYS,\n",
")\n",
"\n",
"# Worker ports\n",
"PORTS = [8002]\n",
"\n",
"# Check workers are running\n",
"healthy = check_workers(PORTS, \"EasyOCR\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "tune",
"metadata": {},
"outputs": [],
"source": [
"# Create trainable and run tuning\n",
"trainable = create_trainable(PORTS, easyocr_payload)\n",
"\n",
"results = run_tuner(\n",
" trainable=trainable,\n",
" search_space=EASYOCR_SEARCH_SPACE,\n",
" num_samples=64,\n",
" num_workers=len(healthy),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "analysis",
"metadata": {},
"outputs": [],
"source": [
"# Analyze results\n",
"df = analyze_results(\n",
" results,\n",
" prefix=\"raytune_easyocr\",\n",
" config_keys=EASYOCR_CONFIG_KEYS,\n",
")\n",
"\n",
"df.describe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "correlation",
"metadata": {},
"outputs": [],
"source": [
"# Correlation analysis\n",
"correlation_analysis(df, EASYOCR_CONFIG_KEYS)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"name": "python",
"version": "3.10.0"
}
},
"nbformat": 4,
"nbformat_minor": 5
}

29
src/easyocr_service/dataset_manager.py
View File

@@ -43,3 +43,32 @@ class ImageTextDataset:
text = f.read()
return image, text
def get_output_path(self, idx, output_subdir, debugset_root="/app/debugset"):
"""Get output path for saving OCR result to debugset folder.
Args:
idx: Sample index
output_subdir: Subdirectory name (e.g., 'paddle_text', 'doctr_text')
debugset_root: Root folder for debug output (default: /app/debugset)
Returns:
Path like /app/debugset/doc1/{output_subdir}/page_001.txt
"""
img_path, _ = self.samples[idx]
# img_path: /app/dataset/doc1/img/page_001.png
# Extract relative path: doc1/img/page_001.png
parts = img_path.split("/dataset/", 1)
if len(parts) == 2:
rel_path = parts[1] # doc1/img/page_001.png
else:
rel_path = os.path.basename(img_path)
# Replace /img/ with /{output_subdir}/
rel_parts = rel_path.rsplit("/img/", 1)
doc_folder = rel_parts[0] # doc1
fname = os.path.splitext(rel_parts[1])[0] + ".txt" # page_001.txt
out_dir = os.path.join(debugset_root, doc_folder, output_subdir)
os.makedirs(out_dir, exist_ok=True)
return os.path.join(out_dir, fname)

2
src/easyocr_service/docker-compose.yml
View File

@@ -14,6 +14,7 @@ services:
- "8002:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- easyocr-cache:/root/.EasyOCR
environment:
- PYTHONUNBUFFERED=1
@@ -34,6 +35,7 @@ services:
- "8002:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- easyocr-cache:/root/.EasyOCR
environment:
- PYTHONUNBUFFERED=1

7
src/easyocr_service/easyocr_tuning_rest.py
View File

@@ -133,6 +133,7 @@ class EvaluateRequest(BaseModel):
# Page range
start_page: int = Field(5, ge=0, description="Start page index (inclusive)")
end_page: int = Field(10, ge=1, description="End page index (exclusive)")
save_output: bool = Field(False, description="Save OCR predictions to debugset folder")
class EvaluateResponse(BaseModel):
@@ -301,6 +302,12 @@ def evaluate(request: EvaluateRequest):
pred = assemble_easyocr_result(result)
time_per_page_list.append(float(time.time() - tp0))
# Save prediction to debugset if requested
if request.save_output:
out_path = state.dataset.get_output_path(idx, "easyocr_text")
with open(out_path, "w", encoding="utf-8") as f:
f.write(pred)
m = evaluate_text(ref, pred)
cer_list.append(m["CER"])
wer_list.append(m["WER"])

1037
src/output_raytune.ipynb
View File

File diff suppressed because it is too large Load Diff

29
src/paddle_ocr/dataset_manager.py
View File

@@ -43,3 +43,32 @@ class ImageTextDataset:
text = f.read()
return image, text
def get_output_path(self, idx, output_subdir, debugset_root="/app/debugset"):
"""Get output path for saving OCR result to debugset folder.
Args:
idx: Sample index
output_subdir: Subdirectory name (e.g., 'paddle_text', 'doctr_text')
debugset_root: Root folder for debug output (default: /app/debugset)
Returns:
Path like /app/debugset/doc1/{output_subdir}/page_001.txt
"""
img_path, _ = self.samples[idx]
# img_path: /app/dataset/doc1/img/page_001.png
# Extract relative path: doc1/img/page_001.png
parts = img_path.split("/dataset/", 1)
if len(parts) == 2:
rel_path = parts[1] # doc1/img/page_001.png
else:
rel_path = os.path.basename(img_path)
# Replace /img/ with /{output_subdir}/
rel_parts = rel_path.rsplit("/img/", 1)
doc_folder = rel_parts[0] # doc1
fname = os.path.splitext(rel_parts[1])[0] + ".txt" # page_001.txt
out_dir = os.path.join(debugset_root, doc_folder, output_subdir)
os.makedirs(out_dir, exist_ok=True)
return os.path.join(out_dir, fname)

1
src/paddle_ocr/docker-compose.cpu-registry.yml
View File

@@ -9,6 +9,7 @@ services:
- "8001:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- paddlex-cache:/root/.paddlex
environment:
- PYTHONUNBUFFERED=1

1
src/paddle_ocr/docker-compose.gpu-registry.yml
View File

@@ -11,6 +11,7 @@ services:
- "8002:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- paddlex-cache:/root/.paddlex
- ./scripts:/app/scripts:ro
environment:

2
src/paddle_ocr/docker-compose.workers.yml
View File

@@ -16,6 +16,7 @@ x-ocr-gpu-common: &ocr-gpu-common
image: seryus.ddns.net/unir/paddle-ocr-gpu:latest
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- paddlex-cache:/root/.paddlex
environment:
- PYTHONUNBUFFERED=1
@@ -39,6 +40,7 @@ x-ocr-cpu-common: &ocr-cpu-common
image: seryus.ddns.net/unir/paddle-ocr-cpu:latest
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- paddlex-cache:/root/.paddlex
environment:
- PYTHONUNBUFFERED=1

4
src/paddle_ocr/docker-compose.yml
View File

@@ -45,7 +45,8 @@ services:
ports:
- "8000:8000"
volumes:
- ../dataset:/app/dataset:ro # Your dataset
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw # Your dataset
- paddlex-cache:/root/.paddlex # For additional models at runtime
environment:
- PYTHONUNBUFFERED=1
@@ -74,6 +75,7 @@ services:
- "8000:8000"
volumes:
- ../dataset:/app/dataset:ro
- ../debugset:/app/debugset:rw
- paddlex-cache:/root/.paddlex
environment:
- PYTHONUNBUFFERED=1

7
src/paddle_ocr/paddle_ocr_tuning_rest.py
View File

@@ -127,6 +127,7 @@ class EvaluateRequest(BaseModel):
text_rec_score_thresh: float = Field(0.0, ge=0.0, le=1.0, description="Recognition score threshold")
start_page: int = Field(5, ge=0, description="Start page index (inclusive)")
end_page: int = Field(10, ge=1, description="End page index (exclusive)")
save_output: bool = Field(False, description="Save OCR predictions to debugset folder")
class EvaluateResponse(BaseModel):
@@ -307,6 +308,12 @@ def evaluate(request: EvaluateRequest):
pred = assemble_from_paddle_result(out)
time_per_page_list.append(float(time.time() - tp0))
# Save prediction to debugset if requested
if request.save_output:
out_path = state.dataset.get_output_path(idx, "paddle_text")
with open(out_path, "w", encoding="utf-8") as f:
f.write(pred)
m = evaluate_text(ref, pred)
cer_list.append(m["CER"])
wer_list.append(m["WER"])

279
src/paddle_ocr_raytune_rest.ipynb
View File

@@ -7,263 +7,81 @@
"source": [
"# PaddleOCR Hyperparameter Optimization via REST API\n",
"\n",
"This notebook runs Ray Tune hyperparameter search calling the PaddleOCR REST API (Docker container).\n",
"Uses Ray Tune + Optuna to find optimal PaddleOCR parameters.\n",
"\n",
"**Benefits:**\n",
"- No model reload per trial - Model stays loaded in Docker container\n",
"- Faster trials - Skip ~10s model load time per trial\n",
"- Cleaner code - REST API replaces subprocess + CLI arg parsing"
]
},
{
"cell_type": "markdown",
"id": "prereq",
"metadata": {},
"source": [
"## Prerequisites\n",
"\n",
"Start 2 PaddleOCR workers for parallel hyperparameter tuning:\n",
"\n",
"```bash\n",
"cd src/paddle_ocr\n",
"docker compose -f docker-compose.workers.yml up\n",
"```\n",
"\n",
"This starts 2 GPU workers on ports 8001-8002, allowing 2 concurrent trials.\n",
"\n",
"For CPU-only systems:\n",
"```bash\n",
"docker compose -f docker-compose.workers.yml --profile cpu up\n",
"docker compose -f docker-compose.workers.yml up # GPU workers on 8001-8002\n",
"# or: docker compose -f docker-compose.workers.yml --profile cpu up\n",
"```"
]
},
{
"cell_type": "markdown",
"id": "3ob9fsoilc4",
"cell_type": "code",
"execution_count": null,
"id": "deps",
"metadata": {},
"outputs": [],
"source": [
"## 0. Dependencies"
"%pip install -q -U \"ray[tune]\" optuna requests pandas"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "wyr2nsoj7",
"id": "setup",
"metadata": {},
"outputs": [],
"source": [
"# Install dependencies (run once)\n",
"%pip install -U \"ray[tune]\"\n",
"%pip install optuna\n",
"%pip install requests pandas"
]
},
{
"cell_type": "markdown",
"id": "imports-header",
"metadata": {},
"source": [
"## 1. Imports & Setup"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "imports",
"metadata": {},
"outputs": [],
"source": "import os\nfrom datetime import datetime\n\nimport requests\nimport pandas as pd\n\nimport ray\nfrom ray import tune, train\nfrom ray.tune.search.optuna import OptunaSearch"
},
{
"cell_type": "markdown",
"id": "config-header",
"metadata": {},
"source": [
"## 2. API Configuration"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "config",
"metadata": {},
"outputs": [],
"source": [
"# PaddleOCR REST API endpoints - 2 workers for parallel trials\n",
"# Start workers with: cd src/paddle_ocr && docker compose -f docker-compose.workers.yml up\n",
"WORKER_PORTS = [8001, 8002]\n",
"WORKER_URLS = [f\"http://localhost:{port}\" for port in WORKER_PORTS]\n",
"\n",
"# Output folder for results\n",
"OUTPUT_FOLDER = \"results\"\n",
"os.makedirs(OUTPUT_FOLDER, exist_ok=True)\n",
"\n",
"# Number of concurrent trials = number of workers\n",
"NUM_WORKERS = len(WORKER_URLS)"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "health-check",
"metadata": {},
"outputs": [],
"source": [
"# Verify all workers are running\n",
"healthy_workers = []\n",
"for url in WORKER_URLS:\n",
" try:\n",
" health = requests.get(f\"{url}/health\", timeout=10).json()\n",
" if health['status'] == 'ok' and health['model_loaded']:\n",
" healthy_workers.append(url)\n",
" print(f\"✓ {url}: {health['status']} (GPU: {health.get('gpu_name', 'N/A')})\")\n",
" else:\n",
" print(f\"✗ {url}: not ready yet\")\n",
" except requests.exceptions.ConnectionError:\n",
" print(f\"✗ {url}: not reachable\")\n",
"\n",
"if not healthy_workers:\n",
" raise RuntimeError(\n",
" \"No healthy workers found. Start them with:\\n\"\n",
" \" cd src/paddle_ocr && docker compose -f docker-compose.workers.yml up\"\n",
"from raytune_ocr import (\n",
" check_workers, create_trainable, run_tuner, analyze_results, correlation_analysis,\n",
" paddle_ocr_payload, PADDLE_OCR_SEARCH_SPACE, PADDLE_OCR_CONFIG_KEYS,\n",
")\n",
"\n",
"print(f\"\\n{len(healthy_workers)}/{len(WORKER_URLS)} workers ready for parallel tuning\")"
]
},
{
"cell_type": "markdown",
"id": "search-space-header",
"metadata": {},
"source": [
"## 3. Search Space"
"# Worker ports\n",
"PORTS = [8001, 8002]\n",
"\n",
"# Check workers are running\n",
"healthy = check_workers(PORTS, \"PaddleOCR\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "search-space",
"id": "tune",
"metadata": {},
"outputs": [],
"source": [
"search_space = {\n",
" # Whether to use document image orientation classification\n",
" \"use_doc_orientation_classify\": tune.choice([True, False]),\n",
" # Whether to use text image unwarping\n",
" \"use_doc_unwarping\": tune.choice([True, False]),\n",
" # Whether to use text line orientation classification\n",
" \"textline_orientation\": tune.choice([True, False]),\n",
" # Detection pixel threshold (pixels > threshold are considered text)\n",
" \"text_det_thresh\": tune.uniform(0.0, 0.7),\n",
" # Detection box threshold (average score within border)\n",
" \"text_det_box_thresh\": tune.uniform(0.0, 0.7),\n",
" # Text detection expansion coefficient\n",
" \"text_det_unclip_ratio\": tune.choice([0.0]),\n",
" # Text recognition threshold (filter low confidence results)\n",
" \"text_rec_score_thresh\": tune.uniform(0.0, 0.7),\n",
"}"
]
},
{
"cell_type": "markdown",
"id": "trainable-header",
"metadata": {},
"source": [
"## 4. Trainable Function"
"# Create trainable and run tuning\n",
"trainable = create_trainable(PORTS, paddle_ocr_payload)\n",
"\n",
"results = run_tuner(\n",
" trainable=trainable,\n",
" search_space=PADDLE_OCR_SEARCH_SPACE,\n",
" num_samples=64,\n",
" num_workers=len(healthy),\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "trainable",
"metadata": {},
"outputs": [],
"source": "def trainable_paddle_ocr(config):\n \"\"\"Call PaddleOCR REST API with the given hyperparameter config.\"\"\"\n import random\n import requests\n from ray import train\n\n # Worker URLs - random selection (load balances with 2 workers, 2 concurrent trials)\n WORKER_PORTS = [8001, 8002]\n api_url = f\"http://localhost:{random.choice(WORKER_PORTS)}\"\n\n payload = {\n \"pdf_folder\": \"/app/dataset\",\n \"use_doc_orientation_classify\": config.get(\"use_doc_orientation_classify\", False),\n \"use_doc_unwarping\": config.get(\"use_doc_unwarping\", False),\n \"textline_orientation\": config.get(\"textline_orientation\", True),\n \"text_det_thresh\": config.get(\"text_det_thresh\", 0.0),\n \"text_det_box_thresh\": config.get(\"text_det_box_thresh\", 0.0),\n \"text_det_unclip_ratio\": config.get(\"text_det_unclip_ratio\", 1.5),\n \"text_rec_score_thresh\": config.get(\"text_rec_score_thresh\", 0.0),\n \"start_page\": 5,\n \"end_page\": 10,\n }\n\n try:\n response = requests.post(f\"{api_url}/evaluate\", json=payload, timeout=None)\n response.raise_for_status()\n metrics = response.json()\n metrics[\"worker\"] = api_url\n train.report(metrics)\n except Exception as e:\n train.report({\n \"CER\": 1.0,\n \"WER\": 1.0,\n \"TIME\": 0.0,\n \"PAGES\": 0,\n \"TIME_PER_PAGE\": 0,\n \"worker\": api_url,\n \"ERROR\": str(e)[:500]\n })"
},
{
"cell_type": "markdown",
"id": "tuner-header",
"metadata": {},
"source": [
"## 5. Run Tuner"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "ray-init",
"id": "analysis",
"metadata": {},
"outputs": [],
"source": [
"ray.init(ignore_reinit_error=True)\n",
"print(f\"Ray Tune ready (version: {ray.__version__})\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "tuner",
"metadata": {},
"outputs": [],
"source": "tuner = tune.Tuner(\n trainable_paddle_ocr,\n tune_config=tune.TuneConfig(\n metric=\"CER\",\n mode=\"min\",\n search_alg=OptunaSearch(),\n num_samples=64,\n max_concurrent_trials=NUM_WORKERS, # Run trials in parallel across workers\n ),\n param_space=search_space,\n)\n\nresults = tuner.fit()"
},
{
"cell_type": "markdown",
"id": "analysis-header",
"metadata": {},
"source": [
"## 6. Results Analysis"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "results-df",
"metadata": {},
"outputs": [],
"source": [
"df = results.get_dataframe()\n",
"# Analyze results\n",
"df = analyze_results(\n",
" results,\n",
" prefix=\"raytune_paddle\",\n",
" config_keys=PADDLE_OCR_CONFIG_KEYS,\n",
")\n",
"\n",
"df.describe()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "save-results",
"metadata": {},
"outputs": [],
"source": [
"# Save results to CSV\n",
"timestamp = datetime.now().strftime(\"%Y%m%d_%H%M%S\")\n",
"filename = f\"raytune_paddle_rest_results_{timestamp}.csv\"\n",
"filepath = os.path.join(OUTPUT_FOLDER, filename)\n",
"\n",
"df.to_csv(filepath, index=False)\n",
"print(f\"Results saved: {filepath}\")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "best-config",
"metadata": {},
"outputs": [],
"source": [
"# Best configuration\n",
"best = df.loc[df[\"CER\"].idxmin()]\n",
"\n",
"print(f\"Best CER: {best['CER']:.6f}\")\n",
"print(f\"Best WER: {best['WER']:.6f}\")\n",
"print(f\"\\nOptimal Configuration:\")\n",
"print(f\" textline_orientation: {best['config/textline_orientation']}\")\n",
"print(f\" use_doc_orientation_classify: {best['config/use_doc_orientation_classify']}\")\n",
"print(f\" use_doc_unwarping: {best['config/use_doc_unwarping']}\")\n",
"print(f\" text_det_thresh: {best['config/text_det_thresh']:.4f}\")\n",
"print(f\" text_det_box_thresh: {best['config/text_det_box_thresh']:.4f}\")\n",
"print(f\" text_det_unclip_ratio: {best['config/text_det_unclip_ratio']}\")\n",
"print(f\" text_rec_score_thresh: {best['config/text_rec_score_thresh']:.4f}\")"
]
},
{
"cell_type": "code",
"execution_count": null,
@@ -272,40 +90,19 @@
"outputs": [],
"source": [
"# Correlation analysis\n",
"param_cols = [\n",
" \"config/text_det_thresh\",\n",
" \"config/text_det_box_thresh\",\n",
" \"config/text_det_unclip_ratio\",\n",
" \"config/text_rec_score_thresh\",\n",
"]\n",
"\n",
"corr_cer = df[param_cols + [\"CER\"]].corr()[\"CER\"].sort_values(ascending=False)\n",
"corr_wer = df[param_cols + [\"WER\"]].corr()[\"WER\"].sort_values(ascending=False)\n",
"\n",
"print(\"Correlation with CER:\")\n",
"print(corr_cer)\n",
"print(\"\\nCorrelation with WER:\")\n",
"print(corr_wer)"
"correlation_analysis(df, PADDLE_OCR_CONFIG_KEYS)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": ".venv",
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.12.3"
"version": "3.10.0"
}
},
"nbformat": 4,

333
src/raytune_ocr.py Normal file
View File

@@ -0,0 +1,333 @@
# raytune_ocr.py
# Shared Ray Tune utilities for OCR hyperparameter optimization
#
# Usage:
# from raytune_ocr import check_workers, create_trainable, run_tuner, analyze_results
import os
from datetime import datetime
from typing import List, Dict, Any, Callable
import requests
import pandas as pd
import ray
from ray import tune, train
from ray.tune.search.optuna import OptunaSearch
def check_workers(ports: List[int], service_name: str = "OCR") -> List[str]:
"""
Verify workers are running and return healthy URLs.
Args:
ports: List of port numbers to check
service_name: Name for error messages
Returns:
List of healthy worker URLs
Raises:
RuntimeError if no healthy workers found
"""
worker_urls = [f"http://localhost:{port}" for port in ports]
healthy_workers = []
for url in worker_urls:
try:
health = requests.get(f"{url}/health", timeout=10).json()
if health.get('status') == 'ok' and health.get('model_loaded'):
healthy_workers.append(url)
gpu = health.get('gpu_name', 'CPU')
print(f"✓ {url}: {health['status']} ({gpu})")
else:
print(f"✗ {url}: not ready yet")
except requests.exceptions.ConnectionError:
print(f"✗ {url}: not reachable")
if not healthy_workers:
raise RuntimeError(
f"No healthy {service_name} workers found.\n"
f"Checked ports: {ports}"
)
print(f"\n{len(healthy_workers)}/{len(worker_urls)} workers ready")
return healthy_workers
def create_trainable(ports: List[int], payload_fn: Callable[[Dict], Dict]) -> Callable:
"""
Factory to create a trainable function for Ray Tune.
Args:
ports: List of worker ports for load balancing
payload_fn: Function that takes config dict and returns API payload dict
Returns:
Trainable function for Ray Tune
"""
def trainable(config):
import random
import requests
from ray import train
api_url = f"http://localhost:{random.choice(ports)}"
payload = payload_fn(config)
try:
response = requests.post(f"{api_url}/evaluate", json=payload, timeout=None)
response.raise_for_status()
metrics = response.json()
metrics["worker"] = api_url
train.report(metrics)
except Exception as e:
train.report({
"CER": 1.0,
"WER": 1.0,
"TIME": 0.0,
"PAGES": 0,
"TIME_PER_PAGE": 0,
"worker": api_url,
"ERROR": str(e)[:500]
})
return trainable
def run_tuner(
trainable: Callable,
search_space: Dict[str, Any],
num_samples: int = 64,
num_workers: int = 1,
metric: str = "CER",
mode: str = "min",
) -> tune.ResultGrid:
"""
Initialize Ray and run hyperparameter tuning.
Args:
trainable: Trainable function from create_trainable()
search_space: Dict of parameter names to tune.* search spaces
num_samples: Number of trials to run
num_workers: Max concurrent trials
metric: Metric to optimize
mode: "min" or "max"
Returns:
Ray Tune ResultGrid
"""
ray.init(ignore_reinit_error=True, include_dashboard=False)
print(f"Ray Tune ready (version: {ray.__version__})")
tuner = tune.Tuner(
trainable,
tune_config=tune.TuneConfig(
metric=metric,
mode=mode,
search_alg=OptunaSearch(),
num_samples=num_samples,
max_concurrent_trials=num_workers,
),
param_space=search_space,
)
return tuner.fit()
def analyze_results(
results: tune.ResultGrid,
output_folder: str = "results",
prefix: str = "raytune",
config_keys: List[str] = None,
) -> pd.DataFrame:
"""
Analyze and save tuning results.
Args:
results: Ray Tune ResultGrid
output_folder: Directory to save CSV
prefix: Filename prefix
config_keys: List of config keys to show in best result (without 'config/' prefix)
Returns:
Results DataFrame
"""
os.makedirs(output_folder, exist_ok=True)
df = results.get_dataframe()
# Save to CSV
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
filename = f"{prefix}_results_{timestamp}.csv"
filepath = os.path.join(output_folder, filename)
df.to_csv(filepath, index=False)
print(f"Results saved: {filepath}")
# Best configuration
best = df.loc[df["CER"].idxmin()]
print(f"\nBest CER: {best['CER']:.6f}")
print(f"Best WER: {best['WER']:.6f}")
if config_keys:
print(f"\nOptimal Configuration:")
for key in config_keys:
col = f"config/{key}"
if col in best:
val = best[col]
if isinstance(val, float):
print(f" {key}: {val:.4f}")
else:
print(f" {key}: {val}")
return df
def correlation_analysis(df: pd.DataFrame, param_keys: List[str]) -> None:
"""
Print correlation of numeric parameters with CER/WER.
Args:
df: Results DataFrame
param_keys: List of config keys (without 'config/' prefix)
"""
param_cols = [f"config/{k}" for k in param_keys if f"config/{k}" in df.columns]
numeric_cols = [c for c in param_cols if df[c].dtype in ['float64', 'int64']]
if not numeric_cols:
print("No numeric parameters for correlation analysis")
return
corr_cer = df[numeric_cols + ["CER"]].corr()["CER"].sort_values(ascending=False)
corr_wer = df[numeric_cols + ["WER"]].corr()["WER"].sort_values(ascending=False)
print("Correlation with CER:")
print(corr_cer)
print("\nCorrelation with WER:")
print(corr_wer)
# =============================================================================
# OCR-specific payload functions
# =============================================================================
def paddle_ocr_payload(config: Dict, start_page: int = 5, end_page: int = 10, save_output: bool = False) -> Dict:
"""Create payload for PaddleOCR API."""
return {
"pdf_folder": "/app/dataset",
"use_doc_orientation_classify": config.get("use_doc_orientation_classify", False),
"use_doc_unwarping": config.get("use_doc_unwarping", False),
"textline_orientation": config.get("textline_orientation", True),
"text_det_thresh": config.get("text_det_thresh", 0.0),
"text_det_box_thresh": config.get("text_det_box_thresh", 0.0),
"text_det_unclip_ratio": config.get("text_det_unclip_ratio", 1.5),
"text_rec_score_thresh": config.get("text_rec_score_thresh", 0.0),
"start_page": start_page,
"end_page": end_page,
"save_output": save_output,
}
def doctr_payload(config: Dict, start_page: int = 5, end_page: int = 10, save_output: bool = False) -> Dict:
"""Create payload for DocTR API."""
return {
"pdf_folder": "/app/dataset",
"assume_straight_pages": config.get("assume_straight_pages", True),
"straighten_pages": config.get("straighten_pages", False),
"preserve_aspect_ratio": config.get("preserve_aspect_ratio", True),
"symmetric_pad": config.get("symmetric_pad", True),
"disable_page_orientation": config.get("disable_page_orientation", False),
"disable_crop_orientation": config.get("disable_crop_orientation", False),
"resolve_lines": config.get("resolve_lines", True),
"resolve_blocks": config.get("resolve_blocks", False),
"paragraph_break": config.get("paragraph_break", 0.035),
"start_page": start_page,
"end_page": end_page,
"save_output": save_output,
}
def easyocr_payload(config: Dict, start_page: int = 5, end_page: int = 10, save_output: bool = False) -> Dict:
"""Create payload for EasyOCR API."""
return {
"pdf_folder": "/app/dataset",
"text_threshold": config.get("text_threshold", 0.7),
"low_text": config.get("low_text", 0.4),
"link_threshold": config.get("link_threshold", 0.4),
"slope_ths": config.get("slope_ths", 0.1),
"ycenter_ths": config.get("ycenter_ths", 0.5),
"height_ths": config.get("height_ths", 0.5),
"width_ths": config.get("width_ths", 0.5),
"add_margin": config.get("add_margin", 0.1),
"contrast_ths": config.get("contrast_ths", 0.1),
"adjust_contrast": config.get("adjust_contrast", 0.5),
"decoder": config.get("decoder", "greedy"),
"beamWidth": config.get("beamWidth", 5),
"min_size": config.get("min_size", 10),
"start_page": start_page,
"end_page": end_page,
"save_output": save_output,
}
# =============================================================================
# Search spaces
# =============================================================================
PADDLE_OCR_SEARCH_SPACE = {
"use_doc_orientation_classify": tune.choice([True, False]),
"use_doc_unwarping": tune.choice([True, False]),
"textline_orientation": tune.choice([True, False]),
"text_det_thresh": tune.uniform(0.0, 0.7),
"text_det_box_thresh": tune.uniform(0.0, 0.7),
"text_det_unclip_ratio": tune.choice([0.0]),
"text_rec_score_thresh": tune.uniform(0.0, 0.7),
}
DOCTR_SEARCH_SPACE = {
"assume_straight_pages": tune.choice([True, False]),
"straighten_pages": tune.choice([True, False]),
"preserve_aspect_ratio": tune.choice([True, False]),
"symmetric_pad": tune.choice([True, False]),
"disable_page_orientation": tune.choice([True, False]),
"disable_crop_orientation": tune.choice([True, False]),
"resolve_lines": tune.choice([True, False]),
"resolve_blocks": tune.choice([True, False]),
"paragraph_break": tune.uniform(0.01, 0.1),
}
EASYOCR_SEARCH_SPACE = {
"text_threshold": tune.uniform(0.3, 0.9),
"low_text": tune.uniform(0.2, 0.6),
"link_threshold": tune.uniform(0.2, 0.6),
"slope_ths": tune.uniform(0.0, 0.3),
"ycenter_ths": tune.uniform(0.3, 1.0),
"height_ths": tune.uniform(0.3, 1.0),
"width_ths": tune.uniform(0.3, 1.0),
"add_margin": tune.uniform(0.0, 0.3),
"contrast_ths": tune.uniform(0.05, 0.3),
"adjust_contrast": tune.uniform(0.3, 0.8),
"decoder": tune.choice(["greedy", "beamsearch"]),
"beamWidth": tune.choice([3, 5, 7, 10]),
"min_size": tune.choice([5, 10, 15, 20]),
}
# =============================================================================
# Config keys for results display
# =============================================================================
PADDLE_OCR_CONFIG_KEYS = [
"use_doc_orientation_classify", "use_doc_unwarping", "textline_orientation",
"text_det_thresh", "text_det_box_thresh", "text_det_unclip_ratio", "text_rec_score_thresh",
]
DOCTR_CONFIG_KEYS = [
"assume_straight_pages", "straighten_pages", "preserve_aspect_ratio", "symmetric_pad",
"disable_page_orientation", "disable_crop_orientation", "resolve_lines", "resolve_blocks",
"paragraph_break",
]
EASYOCR_CONFIG_KEYS = [
"text_threshold", "low_text", "link_threshold", "slope_ths", "ycenter_ths",
"height_ths", "width_ths", "add_margin", "contrast_ths", "adjust_contrast",
"decoder", "beamWidth", "min_size",
]