[Feature]: Test PyLot and Coco-Models for Object-Detection

[okrusch]

Description

• PyLot's models (Tensorflow) for Object-Detection ito see if it would make a good replacement
• Measure processing time and document results (along with images)
• Evaluate similar Coco PyTorch models for comparison
• Evaluate YOLOv8 for comparison
• Summarize findings und create documentation

Definition of Done

• We can tell if PyLot is good fit for our project, or not
• We know which PyLot model would work best
• We know which PyTorch model would work best
• We know if YOLOv8 would work well

[MaxJa4]

Evaluated the following models in a small standalone project.
Used 8 images for each model to check out different scenarios.
Also measured execution time.

1. PyTorch:

• fasterrcnn_mobilenet_v3_large_320_fpn
• fasterrcnn_resnet50_fpn_v2
• retinanet_resnet50_fpn_v2

2. TensorFlow (Pylot):

• faster-rcnn
• ssdlite-mobilenet-v2
• ssd-mobilenet-fpn-640
• ssd-mobilenet-v1
• ssd-mobilenet-v1-fpn

3. YOLOv8

---

Will add experiment documentation with results soon. -> Done in #71