How To Train and Deploy A Custom EfficientDet Object Detection Model Using TensorFlow 2 (GPU) .

How To Train and Deploy A Custom EfficientDet Object Detection Model Using TensorFlow 2 (GPU) In ODHUB To Detect Multiple Objects With A Webcam In Windows.
deep learning labview
labview deep learning
object detection labview
labview object detection
For more information, please visit us at:
www.anscenter.com
email: [e-mail]
In this video, we are going to use ANS Object Detection (ODHUB) software to design an EfficientDet object detection model for multiple object detection problems. Specifically, we will design the AI model to detect poker cards. We use the dataset from Edje Electronics as our example. The dataset is prepared in the data folder inside the project directory. The data include image files and/or XML label files.
Let get started by opening ANS Object Detection software. This software supports both Cloud training engine and Local training engine. In this example, we will use a Local training engine that reuses local hardware such as dedicated GPU cards. This Local engine utilizes Tensorflow 2 as the backend.
First, we need to select the project directory that contains the data folder. ANS software will auto-detect if the data is valid. Importantly, ANS software supports a built-in labeling tool to allow a user to perform a labeling task. This labeling tool helps a user finish the tedious labeling task with ease. It saves the labeling result into an XML format that is compatible with famous deep learning frameworks.
ANS software also helps to verify if the data to be good for training, such as the number of items per category or miss-match labels, and so on...
The label map file will also be generated automatically.
Then, we send the data to the engine to start the training process.
ANS software simplifies training settings for object detection tasks. Users only need to select the model type (fast, accurate, and very accurate) and the number of steps, ODHUB will take care of the rest. We choose the EfficientDet model (accurate) and start the training process. ODHUB will also detect and use GPU cards automatically.
After the training, we can evaluate the trained model by a useful graphical representation results. It allows us to see the model's performance by looking into its mAP value before we can export the production model.
Finally, ODHUB supports the built-in test interface to allow users to test the model performance before final deployment for production. ODHUB supports a native inference engine that takes advantage of the dedicated GPU card to achieve real-time performance. Users can also choose other data to verify the model behavior.
To support how the model works on real production, ODHUB also provides inference using built-in video capture devices such as webcams. Users can select any built-in video capture devices in their system and test how the trained model performs the detection task in real-time.
After the design process, users can decide to use the trained model in actual applications. Depending on the programming environments and system operating systems, ANSCENTER will provide native API to use the trained model.
In this example, we will National Instrument LabVIEW programming environment to do inference for the detection task.
Although ANSCENTER API works well with NI Vision Development Module, it is not required in this example. That how we turn an ordinary webcam into a smart system using ANS Object Detection (ODHUB) software.

[@waleedelbadry4900]

Hello, Thank you for this fantastic video. I sent you a few questions about Arduino ANNHUB and received no answer on the ANSCENTER contact. Thanks

[@ANSCENTER]

Hi Waleed, Thanks for being interested in our products. We will contact you regarding to ANNHUB.

[@asraabid517]

i am trying to train a model for my dataset. but when training is complete, it says evaluation failed. try again. what to do in this case?