Wolfram Neural Net Repository

Immediate Computable Access to Neural Net Models

YOLO V8 Detect Trained on MS-COCO Data

Detect and localize objects in an image

YOLO (You Only Look Once) Version 8 by Ultralytics is the latest version of the YOLO models. Just like its predecessor, YOLO Version 5, YOLO Version 8 is an anchor-free model that was trained with mosaic augmentation. It features the use of new "C2f" blocks, which employ additional dense connections between bottleneck modules. YOLO Version 8 models outperform all models from previous versions at a similar size.

Training Set Information

Model Information

Examples

Resource retrieval

Get the pre-trained net:

In[1]:=
NetModel["YOLO V8 Detect Trained on MS-COCO Data"]
Out[2]=

NetModel parameters

This model consists of a family of individual nets, each identified by a specific parameter combination. Inspect the available parameters:

In[3]:=
NetModel["YOLO V8 Detect Trained on MS-COCO Data", "ParametersInformation"]
Out[4]=

Pick a non-default net by specifying the parameters:

In[5]:=
NetModel[{"YOLO V8 Detect Trained on MS-COCO Data", "Size" -> "L"}]
Out[6]=

Pick a non-default uninitialized net:

In[7]:=
NetModel[{"YOLO V8 Detect Trained on MS-COCO Data", "Size" -> "L"}, "UninitializedEvaluationNet"]
Out[8]=

Evaluation function

Write an evaluation function to scale the result to the input image size and suppress the least probable detections:

In[9]:=
labels = {"person", "bicycle", "car", "motorcycle", "airplane", "bus", "train", "truck", "boat", "traffic light", "fire hydrant", "stop sign", "parking meter", "bench", "bird", "cat", "dog", "horse", "sheep", "cow", "elephant", "bear", "zebra", "giraffe", "backpack", "umbrella", "handbag", "tie", "suitcase", "frisbee", "skis", "snowboard", "sports ball", "kite", "baseball bat", "baseball glove", "skateboard", "surfboard", "tennis racket", "bottle", "wine glass", "cup", "fork", "knife", "spoon", "bowl", "banana", "apple", "sandwich", "orange", "broccoli", "carrot", "hot dog", "pizza", "donut", "cake", "chair", "couch", "potted plant", "bed", "dining table", "toilet", "tv", "laptop", "mouse", "remote", "keyboard", "cell phone", "microwave", "oven", "toaster", "sink", "refrigerator", "book", "clock", "vase", "scissors", "teddy bear", "hair drier", "toothbrush"};
In[10]:=
netevaluate[net_, img_, detectionThreshold_ : .25, overlapThreshold_ : .5] := Module[{res, imgSize, isDetection, probableClasses, probableBoxes, h, w, max, scale, padding, nms, result}, (*define image dimensions*) imgSize = 640; {w, h} = ImageDimensions[img]; (*get inference*) res = net[img]; (*filter by probability*) (*very small probability are thresholded*) isDetection = UnitStep[Max /@ res["ClassProb"] - detectionThreshold]; {probableClasses, probableBoxes} = Map[Pick[#, isDetection, 1] &, {res["ClassProb"], res["Boxes"]}]; If[Length[probableBoxes] == 0, Return[{}]]; (*transform coordinates into rectangular boxes*) max = Max[{w, h}]; scale = max/imgSize; padding = imgSize*(1 - {w, h}/max)/2; probableBoxes = Apply[Rectangle[ scale*({#1 - #3/2, imgSize - #2 - #4/2} - padding), scale*({#1 + #3/2, imgSize - #2 + #4/2} - padding)] &, probableBoxes, 2]; (*gather the boxes of the same class and perform non- max suppression*) nms = ResourceFunction["NonMaximumSuppression"][ probableBoxes -> Max /@ probableClasses, "Index", MaxOverlapFraction -> overlapThreshold]; <| "Boxes" -> probableBoxes[[nms]], "Classes" -> labels[[Last@*Ordering /@ Part[probableClasses, nms]]] |> ];

Basic usage

Obtain the detected bounding boxes with their corresponding classes and confidences for a given image:

In[11]:=
(* Evaluate this cell to get the example input *) CloudGet["https://www.wolframcloud.com/obj/8acba9d3-bd1e-433e-8309-df969dd2897d"]
In[12]:=
detection = netevaluate[NetModel["YOLO V8 Detect Trained on MS-COCO Data"], testImage];

The model's output is an Association containing the detected "Boxes" and "Classes":

In[13]:=
Keys[detection]
Out[13]=

The "Boxes" key is a list of Rectangle expressions corresponding to the bounding boxes of the detected objects:

In[14]:=
detection["Boxes"]
Out[14]=

The "Classes" key contains the classes of the detected objects:

In[15]:=
detection["Classes"]
Out[15]=

Visualize the detection:

In[16]:=
Image[HighlightImage[testImage, detection["Boxes"], ImageLabels -> detection["Classes"]]]
Out[16]=

Network result

The network computes eight thousand four hundred bounding boxes and the probability that the object is of any given class:

In[17]:=
res = NetModel["YOLO V8 Detect Trained on MS-COCO Data"][testImage];
In[18]:=
Dimensions /@ res
Out[18]=

Visualize the bounding boxes scaled by their class probabilities:

In[19]:=
imgSize = 640; {w, h} = ImageDimensions[testImage]; max = Max[{w, h}]; scale = max/imgSize; padding = imgSize*(1 - {w, h}/max)/2; rectangles = Apply[Rectangle[ scale*({#1 - #3/2, imgSize - #2 - #4/2} - padding), scale*({#1 + #3/2, imgSize - #2 + #4/2} - padding) ] &, res["Boxes"], 2]; Graphics[ MapThread[ {EdgeForm[Opacity[Total[#1] + .01]], #2} &, {Max /@ res["ClassProb"], rectangles} ], BaseStyle -> {FaceForm[], EdgeForm[{Thin, Black}]} ]
Out[25]=

Visualize all the boxes scaled by the probability that they contain a person:

In[26]:=
idx = Position[labels, "person"][[1, 1]]
Out[26]=
In[27]:=
Graphics[ MapThread[{EdgeForm[Opacity[#1 + .01]], #2} &, {Extract[ res["ClassProb"], {All, idx}], rectangles}], BaseStyle -> {FaceForm[], EdgeForm[{Thin, Black}]} ]
Out[27]=

Superimpose the person predictions on top of the input received by the net:

In[28]:=
HighlightImage[testImage, Graphics[ MapThread[{EdgeForm[{Thickness[#1/100], Opacity[(#1 + .01)/3]}], #2} &, {Extract[ res["ClassProb"], {All, idx}], rectangles}]], BaseStyle -> {FaceForm[], EdgeForm[{Thin, Red}]}]
Out[28]=

Net information

Inspect the number of parameters of all arrays in the net:

In[29]:=
Information[ NetModel[ "YOLO V8 Detect Trained on MS-COCO Data"], "ArraysElementCounts"]
Out[30]=

Obtain the total number of parameters:

In[31]:=
Information[ NetModel[ "YOLO V8 Detect Trained on MS-COCO Data"], "ArraysTotalElementCount"]
Out[32]=

Obtain the layer type counts:

In[33]:=
Information[ NetModel[ "YOLO V8 Detect Trained on MS-COCO Data"], "LayerTypeCounts"]
Out[34]=

Display the summary graphic:

In[35]:=
Information[ NetModel["YOLO V8 Detect Trained on MS-COCO Data"], "SummaryGraphic"]
Out[36]=

Export to ONNX

Export the net to the ONNX format:

In[37]:=
onnxFile = Export[FileNameJoin[{$TemporaryDirectory, "net.onnx"}], NetModel["YOLO V8 Detect Trained on MS-COCO Data"]]
Out[38]=

Get the size of the ONNX file:

In[39]:=
FileByteCount[onnxFile]
Out[39]=

The size is similar to the byte count of the resource object:

In[40]:=
NetModel["YOLO V8 Detect Trained on MS-COCO Data", "ByteCount"]
Out[41]=

Check some metadata of the ONNX model:

In[42]:=
{OpsetVersion, IRVersion} = {Import[onnxFile, "OperatorSetVersion"], Import[onnxFile, "IRVersion"]}
Out[42]=

Import the model back into Wolfram Language. However, the NetEncoder and NetDecoder will be absent because they are not supported by ONNX:

In[43]:=
Import[onnxFile]
Out[43]=

Resource History

  • Date Created: 1 September 2023

Reference