Image classification and Object Detection (Faster RCNN multiple-object detection)

This work presents three examples of using transfer-learning to do image classification and object detection

1. Image Classification and object detection.ipynb

• uses MNIST dataset, which contains image with size (28, 28, 1) and classes of 10 (num 0 to 9)
• classify the main subject in an image
• localize it by drawing bounding boxes around it.

example of drawing a box around a image, (original, with_bounding_ox)

and the final result can be evaluated by the iou parameter.

2. Predicting_bounding_box_Caltech_Bird.ipynb

• uses caltech_birds2010 dataset in the tensorflow_datasets.
• predict the location of bounding box using the MobileNet V2 network and fine-tune the model by moving the top layers. The result:
  

3. faster_RCNN_multiple_object_detection.ipynb

• use the model from the tensorflow_hub, including

1. • ssd + mobilenet V2: small and fast.
2. • FasterRCNN + InceptionResNet V2: high accuracy

• image can be downloaded from the wikimedia or using local image

result:

1. the online image:
   

2. the photo of my office: