YOLO example - annotation application (#180)

* YOLO example - annotation application

* quality

* standardize video outputs to 60FPS minimum

* calculate target FPS for video annotations so durations match source

* don't round adjusted fps

* overwrite original FPS with real time target

* use manually named temporary file

* FPS counter

* fps counter fix

* remove video loader frame dropping

* --target-fps support

* update annotation colors

* remove hard check for target fps

* function name fix

* change all colors to blue

* webcam support

* quality

* add constant FPS for target FPS images_per_sec annotations, overwrite original file when target FPS present

* documentation

* images per sec formatting

Author: bfineran

integrations/ultralytics/README.md

@@ -140,6 +140,21 @@ Note: on new Ubuntu systems, to install `cv2` running `sudo apt-get update && ap
 may be necessary.
 
 
+### Annotation Example
+`annotate.py` is a script for using YOLOv3 sparsified (and not sparsified) YOLOv3 models
+to run inferences on images, videos, or webcam streams. For a full list of options
+`python annotate.py -h`.
+
+To run pruned-quantized YOLOv3 on a local webcam run:
+```bash
+python annotate.py \
+    zoo:cv/detection/yolo_v3-spp/pytorch/ultralytics/coco/pruned_quant-aggressive_94 \
+    --source 0 \
+    --quantized-inputs \
+    --image-shape 416 416
+```
+
+
 ### Benchmarking
 `benchmarking.py` is a script for benchmarking sparsified and quantized YOLOv3
 performance with DeepSparse.  For a full list of options run `python benchmarking.py -h`.