I want to Develop an Object Detection RESTful web service that will take images as payload and return the coordinates of the bounding boxes or the image itself marked with the detected objects.
Note: Tensorflow Object Detection API makes it easy to detect objects by using pre-trained object detection models.
If you prefer to use PyTorch instead of TensorFlow, DETECTRON2 (open source project by Facebook AI under Apache 2.0 License) is very powerful for object detection: https://github.com/facebookresearch/detectron2.
Keras is being used by large number of people.. but python built in can also be used in a smart manner..
I am using the python OpenCV library for image detection.
Faizur Rashid Sir thank you for your helpful suggestion
Maybe this is a good choice for you. https://blog.paperspace.com/how-to-implement-a-yolo-object-detector-in-pytorch/
Obviously the OpenCV & Tensorlfow/Keras methods allow for far more in-depth customisation, but if you are looking for a quick and easy, and relatively simple adaptation of an object detection/recognition, then yolo will get you there faster.
Good quick start guide for yolo v3:
https://pjreddie.com/darknet/yolo/
Dear Antony Smith
You can tell me that, Small objects can be estimated using YoloV3 object detection API?
Thank you for your helpful comment
I think there will be limit or at least accuracy trade-off as objects get smaller.
How small are you talking?
Antony Smith, For example, tiny objects (number digits or letters).
Shadikur Rahman
So you're more specifically looking at doing an OCR?
Then there are quite different training models you would want to look at.
If for example you want Hand-Written characters:
https://github.com/ccaldw2/MNIST-classifier
Or if you are reading from printed documents?
The latter is far easier as the characters are consistent.
I think Yolo will be helpful in detecting location on image & classify the objects , as well as it is helpful in predicting bounding box & class probabilities.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics