# What to do after image.manual **URL:** \ **Category:** Uncategorized\ **Tags:** usage, image\ **Created:** [November 19, 2019, 6:53pm UTC](https://support.prodi.gy/t/what-to-do-after-image-manual/2230 "2019-11-19T18:53:06Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![aoliveirahen](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/aoliveirahen/32/1034_2.png) [@aoliveirahen](https://support.prodi.gy/u/aoliveirahen)\ **Post date:** [November 19, 2019, 6:53pm UTC](https://support.prodi.gy/t/what-to-do-after-image-manual/2230/1 "2019-11-19T18:53:06Z")
Hello! What to do after annotate images using image.manual? Should I create/train a model using the annotations in the dataset? If yes, how can I do that? Thanks in advance! ---
**Author:** ![ines](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/ines/32/3_2.png) [@ines](https://support.prodi.gy/u/ines)\ **Post date:** [November 19, 2019, 10:55pm UTC](https://support.prodi.gy/t/what-to-do-after-image-manual/2230/2 "2019-11-19T22:55:59Z")
> [@aoliveirahen](#): > > Should I create/train a model using the annotations in the dataset? Yes, that's one thing you could be doing with the annotations. The data will give you the image data and the pixel coordinates of the bounding boxes. The implementation you choose and how you train it is up to you. And the specifics of the training process depend on the library you choose. If you're just getting started, I think [fast.ai](https://fast.ai) has some good resources and implementations for different computer vision tasks. Also, here's an example and tutorial for using Prodigy with TensorFlow's object detection API: > [@Integrating Tensorflow's Object Detection API with Prodigy](https://support.prodi.gy/t/integrating-tensorflows-object-detection-api-with-prodigy/1965): > > Hello all! Many of you have been asking about a recipe for image.teach and I am really excited to share my work of integrating [Tensorflow's Object Detection API](https://github.com/tensorflow/models/tree/master/research/object_detection) with Prodigy which, I did during this summer in collab with @honnibal and @ines. You can find the source code in [prodigy-recipes](https://github.com/explosion/prodigy-recipes/tree/master/image/tf_odapi) repo. The support is still experimental and feedbacks are welcome! Basically, the point of this post is to act as a guide for the recipe. NOTE: Since we cannot control how the Tensorflow's Object Detection API… ---
**Author:** ![aoliveirahen](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/aoliveirahen/32/1034_2.png) [@aoliveirahen](https://support.prodi.gy/u/aoliveirahen)\ **Post date:** [November 21, 2019, 12:43pm UTC](https://support.prodi.gy/t/what-to-do-after-image-manual/2230/3 "2019-11-21T12:43:36Z")
Thanks a lot, Ines! I will study it 🙂