added roboflow example #855

robertgshaw2-redhat · 2023-01-07T23:18:28Z

No description provided.

review-notebook-app · 2023-01-07T23:18:32Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

mgoin · 2023-01-09T18:19:05Z

examples/roboflow-logging/README.md

+
+Download the dataset. Your $DATASET_URL should look something like `neural-magic/soccer-player-detection-gjrbw/2`.
+```bash
+!roboflow download $DATASET_URL --format yolov5pytorch


Suggested change

!roboflow download $DATASET_URL --format yolov5pytorch

roboflow download $DATASET_URL --format yolov5pytorch

dbogunowicz

lgtm (from the logging code side, I did not dive deep into the technicalities of RoboFlow API). I would clean up the jupyter notebook though (a lot of ugly prints there)

Beth-Kosis · 2023-01-24T19:38:52Z

examples/roboflow-logging/README.md

+5. Re-deploy the newly trained YOLOv5-s model with DeepSparse
+
+Run the following to install the packages needed for the Tutorial. We recommend you
+use a virtual enviornment with Python.


Suggested change

use a virtual enviornment with Python.

use a virtual environment with Python.

examples/roboflow-logging/README.md

This should be an em dash without spaces.

Beth-Kosis

Line 64: Is ORT commonly known? If not, the acronym should be spelled out.

Lines 209-210: Something seems to be missing in the second part of this sentence. But, I can't figure it out. Is it really saying, "...passing the arguments to the constructor" the local path part as a qualifier?
In the loggers section, we declare the RoboflowLogger, passing a local path its definition in roboflow-logger.py the arguments to the constructor (dataset_name and api_key).

Beth-Kosis · 2023-01-25T16:18:46Z

examples/roboflow-logging/README.md

+
+## **What Is Continual Learning?**
+
+The distribution of the incoming data to your production model will vary over time as the enviornment evolves. For example input images to a YOLOv5 model will look very different on a sunny day versus on a rainy day.


Suggested change

The distribution of the incoming data to your production model will vary over time as the enviornment evolves. For example input images to a YOLOv5 model will look very different on a sunny day versus on a rainy day.

The distribution of the incoming data to your production model will vary over time as the environment evolves. For example, input images to a YOLOv5 model will look very different on a sunny day versus on a rainy day.

Beth-Kosis · 2023-01-25T16:21:55Z

examples/roboflow-logging/README.md

+
+## **Tutorial Overview**
+
+This example will demonstrate how to setup a Continual Learning system with Neural Magic and Roboflow


Suggested change

This example will demonstrate how to setup a Continual Learning system with Neural Magic and Roboflow

This example will demonstrate how to set up a Continual Learning system with Neural Magic and Roboflow

Beth-Kosis · 2023-01-25T16:22:22Z

examples/roboflow-logging/README.md

+to train YOLOv5-s to detect soccer players in video clips.
+
+There are 5 steps:
+1. Deploy a YOLOv5-s model trained on COCO with DeepSparse


Suggested change

1. Deploy a YOLOv5-s model trained on COCO with DeepSparse

1. Deploy a YOLOv5-s model trained on COCO with DeepSparse.

Beth-Kosis · 2023-01-25T16:22:34Z

examples/roboflow-logging/README.md

+
+There are 5 steps:
+1. Deploy a YOLOv5-s model trained on COCO with DeepSparse
+2. Log raw images from production to Roboflow


Suggested change

2. Log raw images from production to Roboflow

2. Log raw images from production to Roboflow.

Beth-Kosis · 2023-01-25T16:22:45Z

examples/roboflow-logging/README.md

+There are 5 steps:
+1. Deploy a YOLOv5-s model trained on COCO with DeepSparse
+2. Log raw images from production to Roboflow
+3. Label images with Roboflow


Suggested change

3. Label images with Roboflow

3. Label images with Roboflow.

Beth-Kosis · 2023-01-25T17:14:11Z

examples/roboflow-logging/README.md

+
+## **Step 5: Redeploy the New Model**
+
+We will create a new configuration file for the server which contains two endpoints - one with the sparse YOLOv5s


Suggested change

We will create a new configuration file for the server which contains two endpoints - one with the sparse YOLOv5s

We will create a new configuration file for the Server which contains two endpoints—one with the sparse YOLOv5s

Beth-Kosis · 2023-01-25T17:15:16Z

examples/roboflow-logging/README.md

+trained on COCO and the sparse YOLOv5s fine-tuned on the production data.
+
+Note that we provided a `last.onnx` file (which we trained in the manner described above) in the `server` directory.
+You can swap a path to your ONNX file if desired.


Suggested change

You can swap a path to your ONNX file if desired.

You can swap a path to your ONNX file, if desired.

Beth-Kosis · 2023-01-25T17:16:25Z

examples/roboflow-logging/README.md

+
+You have successfully created a data flywheel for continual learning with DeepSparse and Roboflow!
+
+**Interested In Deploying DeepSparse To Production? [Start your 90 day free trial!](https://neuralmagic.com/deepsparse-free-trial/)**


Suggested change

**Interested In Deploying DeepSparse To Production? [Start your 90 day free trial!](https://neuralmagic.com/deepsparse-free-trial/)**

**Interested in deploying DeepSparse to production? [Start your 90 day free trial!](https://neuralmagic.com/deepsparse-free-trial/)**

Beth-Kosis · 2023-01-25T17:19:57Z

examples/roboflow-logging/README.md

+
+#### **Send Request to the Server**
+
+Now that the model endpoint is running, we can send raw images over HTTP and recieve back 


Suggested change

Now that the model endpoint is running, we can send raw images over HTTP and recieve back

Now that the model endpoint is running, we can send raw images over HTTP and receive back

Beth-Kosis · 2023-01-25T17:24:30Z

examples/roboflow-logging/README.md

+### **Train a Sparse YOLOv5 with SparseML**
+
+SparseML is an open-source optimization library which allows you to create sparse models trained on your dataset. One of the workflows enabled by 
+SparseML is called Sparse Transfer Learning, which is similiar to typical transfer learning, except


Suggested change

SparseML is called Sparse Transfer Learning, which is similiar to typical transfer learning, except

SparseML is called Sparse Transfer Learning, which is similar to typical transfer learning, except

jeanniefinks · 2025-05-10T15:30:32Z

Per the main README announcement, DeepSparse is being deprecated by June 2, 2025. Closing the PR as work has been suspended; thank you for the inputs and support!

added roboflow example

f37178b

robertgshaw2-redhat requested a review from mgoin January 7, 2023 23:18

mgoin reviewed Jan 9, 2023

View reviewed changes

dbogunowicz previously approved these changes Jan 11, 2023

View reviewed changes

still blocked on roboflow import cli

6613431

robertgshaw2-redhat dismissed dbogunowicz’s stale review via 6613431 January 24, 2023 00:40

Beth-Kosis reviewed Jan 24, 2023

View reviewed changes

Beth-Kosis reviewed Jan 25, 2023

View reviewed changes

examples/roboflow-logging/README.md Outdated Show resolved Hide resolved

Update examples/roboflow-logging/README.md

c20d343

This should be an em dash without spaces.

Beth-Kosis reviewed Jan 25, 2023

View reviewed changes

jeanniefinks closed this May 10, 2025

jeanniefinks deleted the rs/roboflow-example branch May 29, 2025 23:52

	!roboflow download $DATASET_URL --format yolov5pytorch
	roboflow download $DATASET_URL --format yolov5pytorch

	use a virtual enviornment with Python.
	use a virtual environment with Python.


		## What Is Continual Learning?

		The distribution of the incoming data to your production model will vary over time as the enviornment evolves. For example input images to a YOLOv5 model will look very different on a sunny day versus on a rainy day.


		## Tutorial Overview

		This example will demonstrate how to setup a Continual Learning system with Neural Magic and Roboflow

	1. Deploy a YOLOv5-s model trained on COCO with DeepSparse
	1. Deploy a YOLOv5-s model trained on COCO with DeepSparse.

	2. Log raw images from production to Roboflow
	2. Log raw images from production to Roboflow.


		## Step 5: Redeploy the New Model

		We will create a new configuration file for the server which contains two endpoints - one with the sparse YOLOv5s

	You can swap a path to your ONNX file if desired.
	You can swap a path to your ONNX file, if desired.


		You have successfully created a data flywheel for continual learning with DeepSparse and Roboflow!

		Interested In Deploying DeepSparse To Production? [Start your 90 day free trial!](https://neuralmagic.com/deepsparse-free-trial/)


		#### Send Request to the Server

		Now that the model endpoint is running, we can send raw images over HTTP and recieve back

	SparseML is called Sparse Transfer Learning, which is similiar to typical transfer learning, except
	SparseML is called Sparse Transfer Learning, which is similar to typical transfer learning, except

added roboflow example #855

added roboflow example #855

Uh oh!

Conversation

robertgshaw2-redhat commented Jan 7, 2023

Uh oh!

review-notebook-app bot commented Jan 7, 2023

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dbogunowicz left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Beth-Kosis left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jeanniefinks commented May 10, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

dbogunowicz left a comment •

edited

Loading