Remove scene detection code (#377)

* Remove scene detection code

* Remove tensorflow from runner

.github/runner/Dockerfile

@@ -37,11 +37,6 @@ RUN cd /home/devops && mkdir actions-runner && cd actions-runner \
     && curl -O -L https://github.com/actions/runner/releases/download/v${RUNNER_VERSION}/actions-runner-linux-x64-${RUNNER_VERSION}.tar.gz \
     && tar xzf ./actions-runner-linux-x64-${RUNNER_VERSION}.tar.gz && chown -R devops ~devops
 
-RUN LIBTENSORFLOW_VERSION=2.6.3 \
-  && curl -LO https://storage.googleapis.com/tensorflow/libtensorflow/libtensorflow-gpu-linux-x86_64-${LIBTENSORFLOW_VERSION}.tar.gz \
-  && sudo tar -C /usr/local -xzf libtensorflow-gpu-linux-x86_64-${LIBTENSORFLOW_VERSION}.tar.gz \
-  && sudo ldconfig
-
 # Add mime type for ts
 RUN sudo echo '<?xml version="1.0" encoding="UTF-8"?><mime-info xmlns="http://www.freedesktop.org/standards/shared-mime-info"><mime-type type="video/mp2t"><comment>ts</comment><glob pattern="*.ts"/></mime-type></mime-info>'>>/usr/share/mime/packages/custom_mime_type.xml
 RUN sudo update-mime-database /usr/share/mime

cmd/transcoding/transcoding.go

@@ -64,9 +64,7 @@ func main() {
 			o := ffmpeg.TranscodeOptions{
 				Oname:   fmt.Sprintf("out_%s_%d_out.mp4", lbl, i),
 				Profile: profs[i],
-				// Uncomment the following to test scene classifier
-				// Detector: &ffmpeg.DSceneAdultSoccer,
-				Accel: accel,
+				Accel:   accel,
 			}
 			o.From = *from
 			o.To = *to
@@ -99,11 +97,7 @@ func main() {
 	end := time.Now()
 	fmt.Printf("profile=input frames=%v pixels=%v\n", res.Decoded.Frames, res.Decoded.Pixels)
 	for i, r := range res.Encoded {
-		if r.DetectData != nil {
-			fmt.Printf("profile=%v frames=%v pixels=%v detectdata=%v\n", profiles[i].Name, r.Frames, r.Pixels, r.DetectData)
-		} else {
-			fmt.Printf("profile=%v frames=%v pixels=%v\n", profiles[i].Name, r.Frames, r.Pixels)
-		}
+		fmt.Printf("profile=%v frames=%v pixels=%v\n", profiles[i].Name, r.Frames, r.Pixels)
 	}
 	fmt.Printf("Transcoding time %0.4v\n", end.Sub(t).Seconds())
 }

ffmpeg/api_test.go

@@ -1520,61 +1520,6 @@ func TestTranscoder_CompareVideo(t *testing.T) {
 	compareVideo(t, Software)
 }
 
-func detectionFreq(t *testing.T, accel Acceleration, deviceid string) {
-	run, dir := setupTest(t)
-	defer os.RemoveAll(dir)
-	cmd := `
-    # run segmenter and sanity check frame counts . Hardcode for now.
-    ffmpeg -loglevel warning -i "$1"/../transcoder/test.ts -c:a copy -c:v copy -f hls test.m3u8
-    ffprobe -loglevel warning -select_streams v -count_frames -show_streams test0.ts | grep nb_read_frames=120
-    ffprobe -loglevel warning -select_streams v -count_frames -show_streams test1.ts | grep nb_read_frames=120
-    ffprobe -loglevel warning -select_streams v -count_frames -show_streams test2.ts | grep nb_read_frames=120
-    ffprobe -loglevel warning -select_streams v -count_frames -show_streams test3.ts | grep nb_read_frames=120
-  `
-	run(cmd)
-
-	InitFFmpeg()
-	tc, err := NewTranscoderWithDetector(&DSceneAdultSoccer, deviceid)
-	require.NotNil(t, tc, "look for `Failed to load native model` logs above")
-	if err != nil {
-		t.Error(err)
-	} else {
-		defer tc.StopTranscoder()
-		// Test encoding with only seg0 and seg2 under detection
-		prof := P144p30fps16x9
-		for i := 0; i < 4; i++ {
-			in := &TranscodeOptionsIn{
-				Fname: fmt.Sprintf("%s/test%d.ts", dir, i),
-				Accel: accel,
-			}
-			out := []TranscodeOptions{
-				{
-					Oname:   fmt.Sprintf("%s/out%d.ts", dir, i),
-					Profile: prof,
-					Accel:   accel,
-				},
-			}
-			if i%2 == 0 {
-				out = append(out, TranscodeOptions{
-					Detector: &DSceneAdultSoccer,
-					Accel:    accel,
-				})
-			}
-			res, err := tc.Transcode(in, out)
-			if err != nil {
-				t.Error(err)
-			}
-			if i%2 == 0 && (len(res.Encoded) < 2 || res.Encoded[1].DetectData == nil) {
-				t.Error("No detect data returned for detection profile")
-			}
-		}
-	}
-}
-
-func TestTranscoder_DetectionFreq(t *testing.T) {
-	detectionFreq(t, Software, "-1")
-}
-
 func discontinuityAudioSegment(t *testing.T, accel Acceleration) {
 	run, dir := setupTest(t)
 	defer os.RemoveAll(dir)

ffmpeg/encoder.c

@@ -222,11 +222,6 @@ int open_output(struct output_ctx *octx, struct input_ctx *ictx)
 
   // add video encoder if a decoder exists and this output requires one
   if (ictx->vc && needs_decoder(octx->video->name)) {
-    if (octx->dnn_filtergraph && !ictx->vc->hw_frames_ctx) {
-      // swap filtergraph with the pre-initialized DNN filtergraph for SW
-      // for HW we handle it later during filter re-init
-      octx->vf.graph = *octx->dnn_filtergraph;
-    }
     ret = init_video_filters(ictx, octx);
     if (ret < 0) LPMS_ERR(open_output_err, "Unable to open video filter");
 
@@ -435,32 +430,6 @@ int mux(AVPacket *pkt, AVRational tb, struct output_ctx *octx, AVStream *ost)
   return av_interleaved_write_frame(octx->oc, pkt);
 }
 
-static int getmetadatainf(AVFrame *inf, struct output_ctx *octx)
-{
-  if(inf == NULL) return -1;
-  char classinfo[128] = {0,};
-  AVDictionaryEntry *element = NULL;
-  AVDictionary *metadata = inf->metadata;
-
-  if(metadata != NULL) {
-    element = av_dict_get(metadata, LVPDNN_FILTER_META, element, 0);
-    if(element != NULL) {
-      strcpy(classinfo, element->value);
-      if(strlen(classinfo) > 0) {
-        char * token = strtok(classinfo, ",");
-        int cid = 0;
-        while( token != NULL ) {
-            octx->res->probs[cid] += atof(token);
-            token = strtok(NULL, ",");
-            cid++;
-        }
-        octx->res->frames++;
-      }
-    }
-  }
-  return 0;
-}
-
 static int calc_signature(AVFrame *inf, struct output_ctx *octx)
 {
   int ret = 0;
@@ -552,19 +521,11 @@ int process_out(struct input_ctx *ictx, struct output_ctx *octx, AVCodecContext
         octx->next_kf_pts = frame->pts + octx->gop_pts_len;
     }
 
-    if(octx->is_dnn_profile) {
-      ret = getmetadatainf(frame, octx);
-      if(ret == -1 && frame == NULL) {
-        // Return EOF in case of flushing procedure
-        ret = AVERROR_EOF;
-      }
-    } else {
       if(is_video && frame != NULL && octx->sfilters != NULL) {
          ret = calc_signature(frame, octx);
          if(ret < 0) LPMS_WARN("Could not calculate signature value for frame");
       }
       ret = encode(encoder, frame, octx, ost);
-    }
 skip:
     av_frame_unref(frame);
     // For HW we keep the encoder open so will only get EAGAIN.