Object detection with Artificial Intelligence (AI)
Object detection is a subfield of AI. Object detection algorithms identifies and tracks objects in videos or images. In AI, Object detection infers to the ability of a computer software to recognize an object in a video/image scene.
Application of Object detection In AI
- Vehicle detection
- Self-driving cars
- Face recognition/detection
- Counting systemes (crowd, pedestrians, fans etc)
I will give a step by step implementation of an object recognition algorithm.
PHASE 1
Install all the deep learning dependencies such as tensorflow-gpu==1.15.0 which are necessary for the implementation.
- tensorflow-gpu==1.15.0
- keras
- imageai (2.0.2 ) etc
PHASE 2
We used the Yolo-tiny.h5 as the model type for a basketball rim scenario detection and obtained the following results:
PHASE 3
We used the resnet50_coco_best_v2.0.1.h5 as the model type for a car traffic scenario detection and obtained the following results:
PHASE 4
We used the Yolo-tiny.h5 as the model type for a basketball scenario detection and obtained the following results:
The code snippet of the implementation for the test cases are at the following link: https://github.com/Jonathan-WS/Object_detection_with_AI.git
Major breakthroughs in Artificial Intelligence has enhanced the integration and development of deep learning algorithms such as the RetinaNet, SSD, YOLO, R-CNN, fast-RCNN, etc for accurate object detection.
References:
https://developer.ridgerun.com/wiki/index.php?title=GstInference/Supported_architectures/TinyYoloV3
https://imageai.readthedocs.io/en/latest/detection/index.html
https://drive.google.com/drive/folders/1U1qlldNr8yL2nfXFMvrj4EvcMh88pYUV?usp=sharing
https://drive.google.com/file/d/1ZAyNCkuSxSdXqSOkn0O5wXOvh4RcN2L6/view?usp=sharing
https://drive.google.com/file/d/1-1omWo7cQ0OGNh12s0RcvfMfo9VkXbrW/view?usp=sharing