add example for gocv

2025-09-27 04:15:53 +08:00 · 2021-07-05 00:06:28 +08:00
parent 6c40f6e9bb
commit 4f101ef477
6 changed files with 251 additions and 127 deletions
--- a/examples/opencv.go
+++ b/examples/opencv.go
@@ -1,112 +0,0 @@
-// +build gocv
-
-package examples
-
-import (
-	"fmt"
-	"image"
-	"image/color"
-	"io"
-	"log"
-
-	ffmpeg "github.com/u2takey/ffmpeg-go"
-	"gocv.io/x/gocv"
-)
-
-func TestExampleOpenCvFaceDetect(t *testing.T) {
-	ExampleFaceDetection("./sample_data/head-pose-face-detection-male-short.mp4",
-		"./sample_data/haarcascade_frontalface_default.xml")
-}
-
-func readProcess(infileName string, writer io.WriteCloser) <-chan error {
-	log.Println("Starting ffmpeg process1")
-	done := make(chan error)
-	go func() {
-		err := ffmpeg.Input(infileName).
-			Output("pipe:",
-				ffmpeg.KwArgs{
-					"format": "rawvideo", "pix_fmt": "rgb24",
-				}).
-			WithOutput(writer).
-			Run()
-		log.Println("ffmpeg process1 done")
-		_ = writer.Close()
-		done <- err
-		close(done)
-	}()
-	return done
-}
-
-func openCvProcess(xmlFile string, reader io.ReadCloser, w, h int) {
-	// open display window
-	window := gocv.NewWindow("Face Detect")
-	defer window.Close()
-
-	// color for the rect when faces detected
-	blue := color.RGBA{0, 0, 255, 0}
-
-	classifier := gocv.NewCascadeClassifier()
-	defer classifier.Close()
-
-	if !classifier.Load(xmlFile) {
-		fmt.Printf("Error reading cascade file: %v\n", xmlFile)
-		return
-	}
-
-	frameSize := w * h * 3
-	buf := make([]byte, frameSize, frameSize)
-	for {
-		n, err := io.ReadFull(reader, buf)
-		if n == 0 || err == io.EOF {
-			return
-		} else if n != frameSize || err != nil {
-			panic(fmt.Sprintf("read error: %d, %s", n, err))
-		}
-		img, err := gocv.NewMatFromBytes(h, w, gocv.MatTypeCV8UC3, buf)
-		if err != nil {
-			fmt.Println("decode fail", err)
-		}
-		if img.Empty() {
-			continue
-		}
-		img2 := gocv.NewMat()
-		gocv.CvtColor(img, &img2, gocv.ColorBGRToRGB)
-
-		// detect faces
-		rects := classifier.DetectMultiScale(img2)
-		fmt.Printf("found %d faces\n", len(rects))
-
-		// draw a rectangle around each face on the original image,
-		// along with text identifing as "Human"
-		for _, r := range rects {
-			gocv.Rectangle(&img2, r, blue, 3)
-
-			size := gocv.GetTextSize("Human", gocv.FontHersheyPlain, 1.2, 2)
-			pt := image.Pt(r.Min.X+(r.Min.X/2)-(size.X/2), r.Min.Y-2)
-			gocv.PutText(&img2, "Human", pt, gocv.FontHersheyPlain, 1.2, blue, 2)
-		}
-
-		// show the image in the window, and wait 1 millisecond
-		window.IMShow(img2)
-		img.Close()
-		img2.Close()
-		if window.WaitKey(10) >= 0 {
-			break
-		}
-	}
-	return
-}
-
-func ExampleFaceDetection(inputFile, xmlFile string) {
-	w, h := getVideoSize(inputFile)
-	log.Println(w, h)
-
-	pr1, pw1 := io.Pipe()
-	done1 := readProcess(inputFile, pw1)
-	openCvProcess(xmlFile, pr1, w, h)
-	err := <-done1
-	if err != nil {
-		panic(err)
-	}
-	log.Println("Done")
-}
--- a/examples/opencv_test.go
+++ b/examples/opencv_test.go
@@ -0,0 +1,197 @@
+// +build gocv
+
+// uncomment line above for gocv examples
+
+package examples
+
+import (
+	"fmt"
+	"image"
+	"image/color"
+	"io"
+	"log"
+	"testing"
+
+	ffmpeg "github.com/u2takey/ffmpeg-go"
+	"gocv.io/x/gocv"
+)
+
+// TestExampleOpenCvFaceDetect will: take a video as input => use opencv for face detection => draw box and show a window
+// This example depends on gocv and opencv, please refer: https://pkg.go.dev/gocv.io/x/gocv for installation.
+func TestExampleOpenCvFaceDetectWithVideo(t *testing.T) {
+	inputFile := "./sample_data/head-pose-face-detection-male-short.mp4"
+	xmlFile := "./sample_data/haarcascade_frontalface_default.xml"
+
+	w, h := getVideoSize(inputFile)
+	log.Println(w, h)
+
+	pr1, pw1 := io.Pipe()
+	readProcess(inputFile, pw1)
+	openCvProcess(xmlFile, pr1, w, h)
+	log.Println("Done")
+}
+
+func readProcess(infileName string, writer io.WriteCloser) {
+	log.Println("Starting ffmpeg process1")
+	go func() {
+		err := ffmpeg.Input(infileName).
+			Output("pipe:",
+				ffmpeg.KwArgs{
+					"format": "rawvideo", "pix_fmt": "rgb24",
+				}).
+			WithOutput(writer).
+			ErrorToStdOut().
+			Run()
+		log.Println("ffmpeg process1 done")
+		_ = writer.Close()
+		if err != nil {
+			panic(err)
+		}
+	}()
+	return
+}
+
+func openCvProcess(xmlFile string, reader io.ReadCloser, w, h int) {
+	// open display window
+	window := gocv.NewWindow("Face Detect")
+	defer window.Close()
+
+	// color for the rect when faces detected
+	blue := color.RGBA{B: 255}
+
+	classifier := gocv.NewCascadeClassifier()
+	defer classifier.Close()
+
+	if !classifier.Load(xmlFile) {
+		fmt.Printf("Error reading cascade file: %v\n", xmlFile)
+		return
+	}
+
+	frameSize := w * h * 3
+	buf := make([]byte, frameSize, frameSize)
+	for {
+		n, err := io.ReadFull(reader, buf)
+		if n == 0 || err == io.EOF {
+			return
+		} else if n != frameSize || err != nil {
+			panic(fmt.Sprintf("read error: %d, %s", n, err))
+		}
+		img, err := gocv.NewMatFromBytes(h, w, gocv.MatTypeCV8UC3, buf)
+		if err != nil {
+			fmt.Println("decode fail", err)
+		}
+		if img.Empty() {
+			continue
+		}
+		img2 := gocv.NewMat()
+		gocv.CvtColor(img, &img2, gocv.ColorBGRToRGB)
+
+		// detect faces
+		rects := classifier.DetectMultiScale(img2)
+		fmt.Printf("found %d faces\n", len(rects))
+
+		// draw a rectangle around each face on the original image, along with text identifing as "Human"
+		for _, r := range rects {
+			gocv.Rectangle(&img2, r, blue, 3)
+
+			size := gocv.GetTextSize("Human", gocv.FontHersheyPlain, 1.2, 2)
+			pt := image.Pt(r.Min.X+(r.Min.X/2)-(size.X/2), r.Min.Y-2)
+			gocv.PutText(&img2, "Human", pt, gocv.FontHersheyPlain, 1.2, blue, 2)
+		}
+
+		// show the image in the window, and wait 1 millisecond
+		window.IMShow(img2)
+		img.Close()
+		img2.Close()
+		if window.WaitKey(10) >= 0 {
+			break
+		}
+	}
+	return
+}
+
+// TestExampleOpenCvFaceDetectWithCamera will: task stream from webcam => use opencv for face detection => output with ffmpeg
+// This example depends on gocv and opencv, please refer: https://pkg.go.dev/gocv.io/x/gocv for installation.
+func TestExampleOpenCvFaceDetectWithCamera(t *testing.T) {
+	deviceID := "0" // camera device id
+	xmlFile := "./sample_data/haarcascade_frontalface_default.xml"
+
+	webcam, err := gocv.OpenVideoCapture(deviceID)
+	if err != nil {
+		fmt.Printf("error opening video capture device: %v\n", deviceID)
+		return
+	}
+	defer webcam.Close()
+
+	// prepare image matrix
+	img := gocv.NewMat()
+	defer img.Close()
+
+	if ok := webcam.Read(&img); !ok {
+		panic(fmt.Sprintf("Cannot read device %v", deviceID))
+	}
+	fmt.Printf("img: %vX%v\n", img.Cols(), img.Rows())
+
+	pr1, pw1 := io.Pipe()
+	writeProcess("./sample_data/face_detect.mp4", pr1, img.Cols(), img.Rows())
+
+	// color for the rect when faces detected
+	blue := color.RGBA{B: 255}
+
+	// load classifier to recognize faces
+	classifier := gocv.NewCascadeClassifier()
+	defer classifier.Close()
+
+	if !classifier.Load(xmlFile) {
+		fmt.Printf("Error reading cascade file: %v\n", xmlFile)
+		return
+	}
+
+	fmt.Printf("Start reading device: %v\n", deviceID)
+	for i := 0; i < 200; i++ {
+		if ok := webcam.Read(&img); !ok {
+			fmt.Printf("Device closed: %v\n", deviceID)
+			return
+		}
+		if img.Empty() {
+			continue
+		}
+
+		// detect faces
+		rects := classifier.DetectMultiScale(img)
+		fmt.Printf("found %d faces\n", len(rects))
+
+		// draw a rectangle around each face on the original image, along with text identifing as "Human"
+		for _, r := range rects {
+			gocv.Rectangle(&img, r, blue, 3)
+
+			size := gocv.GetTextSize("Human", gocv.FontHersheyPlain, 1.2, 2)
+			pt := image.Pt(r.Min.X+(r.Min.X/2)-(size.X/2), r.Min.Y-2)
+			gocv.PutText(&img, "Human", pt, gocv.FontHersheyPlain, 1.2, blue, 2)
+		}
+		pw1.Write(img.ToBytes())
+	}
+	pw1.Close()
+	log.Println("Done")
+}
+
+func writeProcess(outputFile string, reader io.ReadCloser, w, h int) {
+	log.Println("Starting ffmpeg process1")
+	go func() {
+		err := ffmpeg.Input("pipe:",
+			ffmpeg.KwArgs{"format": "rawvideo",
+				"pix_fmt": "bgr24", "s": fmt.Sprintf("%dx%d", w, h),
+			}).
+			Overlay(ffmpeg.Input("./sample_data/overlay.png"), "").
+			Output(outputFile).
+			WithInput(reader).
+			ErrorToStdOut().
+			OverWriteOutput().
+			Run()
+		log.Println("ffmpeg process1 done")
+		if err != nil {
+			panic(err)
+		}
+		_ = reader.Close()
+	}()
+}
--- a/examples/sample_data/out1.mp4
+++ b/examples/sample_data/out1.mp4
--- a/ffmpeg_test.go
+++ b/ffmpeg_test.go
@@ -84,6 +84,31 @@ func TestSimpleOutputArgs(t *testing.T) {
 		"2", "-vf", "fps=3", "imageFromVideo_%d.jpg"}, cmd.GetArgs())
 }

+func TestAutomaticStreamSelection(t *testing.T) {
+	// example from http://ffmpeg.org/ffmpeg-all.html
+	input := []*Stream{Input("A.avi"), Input("B.mp4")}
+	out1 := Output(input, "out1.mkv")
+	out2 := Output(input, "out2.wav")
+	out3 := Output(input, "out3.mov", KwArgs{"map": "1:a", "c:a": "copy"})
+	cmd := MergeOutputs(out1, out2, out3)
+	printArgs(cmd.GetArgs())
+	printGraph(cmd)
+}
+
+func TestLabeledFiltergraph(t *testing.T) {
+	// example from http://ffmpeg.org/ffmpeg-all.html
+	in1, in2, in3 := Input("A.avi"), Input("B.mp4"), Input("C.mkv")
+	in2Split := in2.Get("v").Hue(KwArgs{"s": 0}).Split()
+	overlay := Filter([]*Stream{in1, in2}, "overlay", nil)
+	aresample := Filter([]*Stream{in1, in2, in3}, "aresample", nil)
+	out1 := Output([]*Stream{in2Split.Get("outv1"), overlay, aresample}, "out1.mp4", KwArgs{"an": ""})
+	out2 := Output([]*Stream{in1, in2, in3}, "out2.mkv")
+	out3 := in2Split.Get("outv2").Output("out3.mkv", KwArgs{"map": "1:a:0"})
+	cmd := MergeOutputs(out1, out2, out3)
+	printArgs(cmd.GetArgs())
+	printGraph(cmd)
+}
+
 func ComplexFilterExample() *Stream {
 	split := Input(TestInputFile1).VFlip().Split()
 	split0, split1 := split.Get("0"), split.Get("1")
@@ -311,6 +336,19 @@ func TestView(t *testing.T) {
 	t.Log(b)
 }

+func printArgs(args []string) {
+	for _, a := range args {
+		fmt.Printf("%s ", a)
+	}
+	fmt.Println()
+}
+
+func printGraph(s *Stream) {
+	fmt.Println()
+	v, _ := s.View(ViewTypeFlowChart)
+	fmt.Println(v)
+}
+
 //func TestAvFoundation(t *testing.T) {
 //	out := Input("default:none", KwArgs{"f": "avfoundation", "framerate": "30"}).
 //		Output("output.mp4", KwArgs{"format": "mp4"}).
--- a/filters.go
+++ b/filters.go
@@ -5,7 +5,7 @@ import (
 	"strconv"
 )

-func AssetType(hasType, expectType string, action string) {
+func AssertType(hasType, expectType string, action string) {
 	if hasType != expectType {
 		panic(fmt.Sprintf("cannot %s on non-%s", action, expectType))
 	}
@@ -20,32 +20,32 @@ func Filter(streamSpec []*Stream, filterName string, args Args, kwArgs ...KwArgs
 }

 func (s *Stream) Filter(filterName string, args Args, kwArgs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "filter")
+	AssertType(s.Type, "FilterableStream", "filter")
 	return Filter([]*Stream{s}, filterName, args, MergeKwArgs(kwArgs))
 }

 func (s *Stream) Split() *Node {
-	AssetType(s.Type, "FilterableStream", "split")
+	AssertType(s.Type, "FilterableStream", "split")
 	return NewFilterNode("split", []*Stream{s}, 1, nil, nil)
 }

 func (s *Stream) ASplit() *Node {
-	AssetType(s.Type, "FilterableStream", "asplit")
+	AssertType(s.Type, "FilterableStream", "asplit")
 	return NewFilterNode("asplit", []*Stream{s}, 1, nil, nil)
 }

 func (s *Stream) SetPts(expr string) *Node {
-	AssetType(s.Type, "FilterableStream", "setpts")
+	AssertType(s.Type, "FilterableStream", "setpts")
 	return NewFilterNode("setpts", []*Stream{s}, 1, []string{expr}, nil)
 }

 func (s *Stream) Trim(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "trim")
+	AssertType(s.Type, "FilterableStream", "trim")
 	return NewFilterNode("trim", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }

 func (s *Stream) Overlay(overlayParentNode *Stream, eofAction string, kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "overlay")
+	AssertType(s.Type, "FilterableStream", "overlay")
 	if eofAction == "" {
 		eofAction = "repeat"
 	}
@@ -55,24 +55,24 @@ func (s *Stream) Overlay(overlayParentNode *Stream, eofAction string, kwargs ...
 }

 func (s *Stream) HFlip(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "hflip")
+	AssertType(s.Type, "FilterableStream", "hflip")
 	return NewFilterNode("hflip", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }

 func (s *Stream) VFlip(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "vflip")
+	AssertType(s.Type, "FilterableStream", "vflip")
 	return NewFilterNode("vflip", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }

 func (s *Stream) Crop(x, y, w, h int, kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "crop")
+	AssertType(s.Type, "FilterableStream", "crop")
 	return NewFilterNode("crop", []*Stream{s}, 1, []string{
 		strconv.Itoa(w), strconv.Itoa(h), strconv.Itoa(x), strconv.Itoa(y),
 	}, MergeKwArgs(kwargs)).Stream("", "")
 }

 func (s *Stream) DrawBox(x, y, w, h int, color string, thickness int, kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "drawbox")
+	AssertType(s.Type, "FilterableStream", "drawbox")
 	args := MergeKwArgs(kwargs)
 	if thickness != 0 {
 		args["t"] = thickness
@@ -83,7 +83,7 @@ func (s *Stream) DrawBox(x, y, w, h int, color string, thickness int, kwargs ...
 }

 func (s *Stream) Drawtext(text string, x, y int, escape bool, kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "drawtext")
+	AssertType(s.Type, "FilterableStream", "drawtext")
 	args := MergeKwArgs(kwargs)
 	if escape {
 		text = fmt.Sprintf("%q", text)
@@ -119,17 +119,17 @@ func (s *Stream) Concat(streams []*Stream, kwargs ...KwArgs) *Stream {
 }

 func (s *Stream) ZoomPan(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "zoompan")
+	AssertType(s.Type, "FilterableStream", "zoompan")
 	return NewFilterNode("zoompan", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }

 func (s *Stream) Hue(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "hue")
+	AssertType(s.Type, "FilterableStream", "hue")
 	return NewFilterNode("hue", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }

 // todo fix this
 func (s *Stream) ColorChannelMixer(kwargs ...KwArgs) *Stream {
-	AssetType(s.Type, "FilterableStream", "colorchannelmixer")
+	AssertType(s.Type, "FilterableStream", "colorchannelmixer")
 	return NewFilterNode("colorchannelmixer", []*Stream{s}, 1, nil, MergeKwArgs(kwargs)).Stream("", "")
 }
--- a/view.go
+++ b/view.go
@@ -81,6 +81,7 @@ func visualizeForMermaidAsFlowChart(s *Stream) (string, error) {
 		next := outGoingMap[node.Hash()]
 		for k, v := range next {
 			for _, nextNode := range v {
+				// todo ignore merged output
 				label := string(k)
 				if label == "" {
 					label = "<>"