https://github.com/aryanjain28/indian-roads
An android application that detects objects normally found on Indian roads like - Car, Bicycle, Cow, Truck, etc. I used the TF-Lite model in this application and downloaded specific categories from the COCO dataset to train upon. The application uses the phone's rear camera and detects in real-time.
https://github.com/aryanjain28/indian-roads
android-application object-detection tensorflow tensorflow-examples tflite
Last synced: 6 months ago
JSON representation
An android application that detects objects normally found on Indian roads like - Car, Bicycle, Cow, Truck, etc. I used the TF-Lite model in this application and downloaded specific categories from the COCO dataset to train upon. The application uses the phone's rear camera and detects in real-time.
- Host: GitHub
- URL: https://github.com/aryanjain28/indian-roads
- Owner: aryanjain28
- Created: 2020-07-20T09:44:32.000Z (almost 6 years ago)
- Default Branch: master
- Last Pushed: 2020-07-20T17:01:41.000Z (almost 6 years ago)
- Last Synced: 2025-02-13T19:17:16.639Z (over 1 year ago)
- Topics: android-application, object-detection, tensorflow, tensorflow-examples, tflite
- Language: Jupyter Notebook
- Homepage:
- Size: 3.13 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Introduction
An android application that detects objects usually found on an indian road.
Images for the specific categories is downloaded from COCO dataset, using a script that I wrote (available in Script folder in repository).
Categories includes 8 objects, they are :
1. Cow
2. Dog
3. Truck
4. Bicycle
5. Truck
6. MotorCycle
7. Person
8. Bus
I trained, model from "ssd_mobilenet_v2_quantized_300x300_coco" tensorflow model's zoo and converted it to TF-Lite in order to use it in mobile.
# Usage
The application simply uses phones rear camera and detects objects in real time
# Screenshots

# References
https://medium.com
https://cocodataset.org/
https://www.tensorflow.org/lite/