feat: super gradients integration (#246)

Add new detection class which allows the user to supply a
super-gradients model with the following values;
1. model_arch supported by super-gradients
2. model_path to the model weights
3. yaml file with a names section containing in order the labels for the
classes to the object detection model
4. callback function which takes an image and a super-gradients model
and returns supervision detections

Wlll be used for AF contract and contracts in the future.

Author: pravin-unstructured

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 0.7.4-dev0
+
+* Allows user to use super-gradients model if they have a callback predict function, a yaml file with names field corresponding to classes and a path to the model weights
+
 ## 0.7.3
 
 * Integration of Chipperv2 and additional Chipper functionality, which includes automatic detection of GPU,

requirements/base.in

@@ -2,7 +2,8 @@ layoutparser[layoutmodels,tesseract]
 python-multipart
 huggingface-hub
 opencv-python!=4.7.0.68
-onnx==1.14.1
+super-gradients
+supervision
 # NOTE(benjamin): Pinned because onnxruntime changed the way quantization is done, and we need to update our code to support it
 onnxruntime<1.16
 # NOTE(alan): Pinned because this is when the most recent module we import appeared