Human & Animal Detection and Industrial OCR System

Author: Het Bhutak
Email: [email protected]

This workspace contains two complete computer vision systems:

1. Human & Animal Detection in Videos
   Detects and classifies humans and animals in videos using deep learning.
2. Industrial/Military OCR System
   Extracts and structures text from images of industrial/military stenciled boxes.

---

Table of Contents

• Project Structure
• Setup & Installation
• Running the Systems
  • A. Human & Animal Detection
  • B. Industrial OCR System
• Data Format
• Outputs
• References

---

Project Structure

.
├── human_animal_detection.py
├── streamlit_app_part_a.py
├── part_a_requirements.txt
├── test_videos/
├── outputs/
├── Human and Animal Detection.v3i.yolov5pytorch/
│   ├── data.yaml
│   ├── train/valid/test/ (images & YOLO labels)
│   └── README.*
├── Part2/
│   ├── ocr_system_part_b.py
│   ├── streamlit_app_part_b.py
│   ├── requirements.txt
│   ├── outputs/
│   └── test_images/
└── industrial_ocr/
    ├── outputs/
    └── test_images/

---

Setup & Installation

1. Clone the Repository

git clone <repo-url>
cd <repo-folder>

2. Install Python Dependencies

For Human & Animal Detection

pip install -r part_a_requirements.txt

For Industrial OCR

pip install -r Part2/requirements.txt

Additional Requirement for OCR

• Tesseract OCR Engine
  • Windows: Download here
  • Linux: sudo apt-get install tesseract-ocr
  • Mac: brew install tesseract

---

Running the Systems

A. Human & Animal Detection

Streamlit Web App

streamlit run streamlit_app_part_a.py

• Upload a video in the web UI.
• Process and view annotated results and statistics.

Command Line (Batch Processing)

python human_animal_detection.py

• Processes all videos in test_videos/.
• Annotated videos saved to outputs/.

---

B. Industrial OCR System

Streamlit Web App

streamlit run Part2/streamlit_app_part_b.py

• Upload an image in the web UI.
• Extracts text, shows structured data, and displays annotated images.

Command Line (Batch Processing)

python Part2/ocr_system_part_b.py

• Processes all images in Part2/test_images/.
• Results saved as JSON and annotated images in Part2/outputs/.

---

Data Format

Human & Animal Detection

• Images: JPEG/PNG in Human and Animal Detection.v3i.yolov5pytorch/train/valid/test/images/
• Labels: YOLOv5 format in train/valid/test/labels/
  • Each line: <class_id> <x_center> <y_center> <width> <height> (normalized)
  • class_id: 0 = Animal, 1 = Human

Industrial OCR

• Images: Place test images in Part2/test_images/
• Outputs: JSON files and annotated images in Part2/outputs/

---

Outputs

• Annotated Videos: outputs/annotated_test.mp4
• OCR Results: Part2/outputs/<image_name>_ocr_result.json
• Annotated OCR Images: Part2/outputs/<image_name>_annotated.jpg
• 
• 

---

Notes

• For best results, use clear, high-resolution videos/images.
• All processing is offline; no data is sent to external servers.
• For any issues, check the console output or Streamlit sidebar tips.

---