Object Detection Use Cases

Deep Learning (DL) has achieved huge success in the last years in the field of Computer Vision. Researchers around the world work in the building and training of new state-of-the-art architectures and models. This requires lots of knowledge, skill, data and computing power, who may be hard to get for the vast majority of DL practitioners. That's the main reason because Transfer Learning becomes a very good approach in order to get introduced in this field, which consists of grabbing some architecture/model weights that have demonstrated their value already in well-known image datasets and build new applications starting from there. In this repository appears some apps with different use cases that take use of TensorFlow's Object Detection Models Zoo that may be of interest for those of who require strong Object Detection capabilities.

Dependencies

It's mandatory to download a Tensorflow detection model and locate it outside the project like this:

├───TensorFlow Object Detection Models
│   └───trained_models
│       ├───faster_rcnn_inception_resnet_v2_atrous_coco_11_06_2017
│       │
│       ├───faster_rcnn_resnet101_coco_11_06_2017
│       │
│       ├───rfcn_resnet101_coco_11_06_2017
│       │
│       ├───ssd_inception_v2_coco_11_06_2017
│       │
│       ├───ssd_mobilenet_v1_coco_11_06_2017
│       │
│       └───...
│
├───TF-Object-Detection-Use-Cases
│   │   object_detection_from_url.py
│   │   object_detection_from_video.py
│   │   object_detection_from_webcam.py
│   │   object_detection_with_tensorboard.py
│   │   people_detection_from_webcam.py
│   │   utils.py
│   │   ...
│   │
│   └───object_detection

Depending on the specs of your environment, you may choose one faster or other with a better mean Average Precision (mAP).

Also, depending on the example you want to run you'll need to download certain libraries or have access to hardware (webcam).

If you're a conda user, you can create an environment from the object_detection_environment.yml file using the Terminal or an Anaconda Prompt for the following steps:

1. Create the environment from the object_detection_environment.yml file:

   conda env create -f object_detection_env.yml

2. Activate the new environment:

   • Windows: activate object_detection_env
   • macOS and Linux: source activate object_detection_env

3. Verify that the new environment was installed correctly:

   conda list

You can also clone the environment through the environment manager of Anaconda Navigator.

Use

Launch any of the .py examples from your favorite IDE or from the command prompt. You'll then see two windows, one with the image that's being captured by the program (video, url, webcam, etc.), and other with the same signal but with the detections (bounding boxes) overlapped into it. Press the 'q' button in the resulting floating window to close the program.