Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/oktopuzSlid/detectorPlacas/llms.txt

Use this file to discover all available pages before exploring further.

This guide walks you through every step needed to get DetectorPlacas running on your machine — from installing the correct Python packages to placing model files in the exact locations the scripts expect. Follow each step in order and your environment will be ready to detect license plates in images, video, or a live webcam feed.

​
System Requirements

  • Python 3.6 or 3.7 — required for compatibility with TensorFlow 1.15
  • pip — for installing Python packages
  • CUDA-capable GPU (optional) — accelerates inference; CPU-only execution is supported
1

Install Python packages

Install TensorFlow 1.15, OpenCV, and NumPy using pip:
pip install tensorflow==1.15
pip install opencv-python
pip install numpy
TensorFlow 2.x is not supported. The detection scripts rely exclusively on TF 1.x APIs including tf.app.flags, tf.Graph(), tf.Session(), tf.GraphDef(), and tf.gfile.GFile(). These APIs were removed in TensorFlow 2.x and will cause import errors if you use a newer version.
2

Install the TensorFlow Object Detection API

All three detection scripts import directly from object_detection.utils:
from object_detection.utils import label_map_util
from object_detection.utils import visualization_utils as vis_util
Clone the TensorFlow Models repository and install the research package:
git clone https://github.com/tensorflow/models.git
cd models/research
pip install .
Alternatively, add the research directory to your PYTHONPATH without installing:
export PYTHONPATH=$PYTHONPATH:/path/to/models/research
3

Arrange project files

Every script calls os.getcwd() and builds all paths relative to that directory. MODEL_NAME = 'inference_graph' is hardcoded in all three detection scripts, so the inference_graph/ folder must sit directly inside your project root.Your project root must be laid out as follows:
project-root/        ← run all scripts from here
├── inference_graph/
│   ├── frozen_inference_graph.pb
│   ├── model.ckpt.data-00000-of-00001
│   ├── model.ckpt.index
│   └── model.ckpt.meta
├── saved_model/
│   └── saved_model.pb
├── checkpoint           ← contains: model_checkpoint_path: "model.ckpt"
├── labelmap.pbtxt       ← 1 class: license_plate
├── Object_detection_image.py
├── Object_detection_video.py
├── Object_detection_webcam.py
└── interfaz.py
4

Create labelmap.pbtxt

All scripts set NUM_CLASSES = 1. Your label map file must define exactly one class — license_plate — in the following format:
item {
  id: 1
  name: 'license_plate'
}
Save this file as labelmap.pbtxt in the project root. The path is constructed at runtime as os.path.join(CWD_PATH, 'labelmap.pbtxt').
5

Verify the setup

Run a quick one-liner to confirm all three critical imports resolve without errors:
python -c "import tensorflow as tf; import cv2; from object_detection.utils import label_map_util; print('Setup OK')"
If you see Setup OK, all dependencies are installed and the Object Detection API is on your path.
Always launch detection scripts from the project root directory. Because every path is computed with os.getcwd(), running a script from a different directory will cause FileNotFoundError when the scripts try to open inference_graph/frozen_inference_graph.pb or labelmap.pbtxt.
The checkpoint file in the repository contains two lines that point TensorFlow’s checkpoint reader to the correct model weights:
model_checkpoint_path: "model.ckpt"
all_model_checkpoint_paths: "model.ckpt"

Build docs developers (and LLMs) love

Get started for freeTalk to us