https://github.com/alankrantas/tf-lite-python-object-objection

Object detection examples using pre-trained models on Tensorflow Lite and OpenCV
https://github.com/alankrantas/tf-lite-python-object-objection

deep-learning efficientdet-lite image-classification image-recognition object-detection opencv opencv-python python ssd-mobilenet yolo yolov5

Last synced: about 1 month ago
JSON representation

Object detection examples using pre-trained models on Tensorflow Lite and OpenCV

• Host: GitHub
• URL: https://github.com/alankrantas/tf-lite-python-object-objection
• Owner: alankrantas
• License: mit
• Created: 2022-09-14T14:50:19.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2023-07-11T20:32:39.000Z (almost 2 years ago)
• Last Synced: 2023-07-11T21:30:26.700Z (almost 2 years ago)
• Topics: deep-learning, efficientdet-lite, image-classification, image-recognition, object-detection, opencv, opencv-python, python, ssd-mobilenet, yolo, yolov5
• Language: Python
• Homepage:
• Size: 13.5 MB
• Stars: 0
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Object Detection Examples With Tensorflow Lite and OpenCV (Python)

Running pre-trained TF Lite models for object detection. You either have to install Tehsorflow or Tensorflow Lite (```tflite_runtime```) and OpenCV (```opencv-python```). These scripts also run a lot faster on a ARM device, for example, a Raspberry Pi 3B or 4B.

There are three models available here (downloaded from Google):

* SSD-MobileNet V1

* EfficientDet-Lite0

* YOLO V5

All three are trained with the COCO dataset (```labelmap.txt``` is the label list). This is mainly a demostration of how to get the possible things as well as their location from the model.

![result](https://github.com/alankrantas/TF-Lite-Python-Object-Objection/blob/main/result.jpg)

```TF_Lite_Object_Detection.py``` can use either SSD or EfficientNet to process a still image, and ```TF_Lite_Object_Detection_Yolo.py``` is the YOLO version. ```TF_Lite_Object_Detection_Live.py``` use live USB cam images with SSD or EfficientNet (press ```q```).