[FFmpeg-devel] [PATCH FFmpeg 2/15] libavfilter/dnn: move existing contain_valid_detection_bbox from openvino backend to dnn_backend_common

[m.kaindl0208 at gmail.com]

Moves the contain_valid_detection_bbox function from the OpenVINO backend to the common backend code, making it available for all DNN backends to use when checking bounding box validity.

Will be used by the Torch backend in an upcoming patch in this series.

Try the new filters using my Github Repo https://github.com/MaximilianKaindl/DeepFFMPEGVideoClassification.

Any Feedback is appreciated!

Signed-off-by: MaximilianKaindl <m.kaindl0208 at gmail.com>
---
 libavfilter/dnn/dnn_backend_common.c   | 38 +++++++++++++++++++++++++
 libavfilter/dnn/dnn_backend_common.h   |  9 ++++++
 libavfilter/dnn/dnn_backend_openvino.c | 39 +-------------------------
 3 files changed, 48 insertions(+), 38 deletions(-)

diff --git a/libavfilter/dnn/dnn_backend_common.c b/libavfilter/dnn/dnn_backend_common.c
index e45eefd14d..3944db35b5 100644
--- a/libavfilter/dnn/dnn_backend_common.c
+++ b/libavfilter/dnn/dnn_backend_common.c
@@ -23,6 +23,7 @@
 
 #include "libavutil/mem.h"
 #include "dnn_backend_common.h"
+#include "libavutil/detection_bbox.h"
 
 #define DNN_ASYNC_SUCCESS (void *)0
 #define DNN_ASYNC_FAIL (void *)-1
@@ -178,3 +179,40 @@ int ff_dnn_fill_gettingoutput_task(TaskItem *task, DNNExecBaseParams *exec_param
 
     return ff_dnn_fill_task(task, exec_params, backend_model, 0, 0);
 }
+
+int ff_dnn_contain_valid_detection_bbox(AVFrame *frame)
+{
+    AVFrameSideData *sd;
+    const AVDetectionBBoxHeader *header;
+    const AVDetectionBBox *bbox;
+
+    sd = av_frame_get_side_data(frame, AV_FRAME_DATA_DETECTION_BBOXES);
+    if (!sd) { // this frame has nothing detected
+        return 0;
+    }
+
+    if (!sd->size) {
+        return 0;
+    }
+
+    header = (const AVDetectionBBoxHeader *)sd->data;
+    if (!header->nb_bboxes) {
+        return 0;
+    }
+
+    for (uint32_t i = 0; i < header->nb_bboxes; i++) {
+        bbox = av_get_detection_bbox(header, i);
+        if (bbox->x < 0 || bbox->w < 0 || bbox->x + bbox->w >= frame->width) {
+            return 0;
+        }
+        if (bbox->y < 0 || bbox->h < 0 || bbox->y + bbox->h >= frame->height) {
+            return 0;
+        }
+
+        if (bbox->classify_count == AV_NUM_DETECTION_BBOX_CLASSIFY) {
+            return 0;
+        }
+    }
+
+    return 1;
+}
\ No newline at end of file
diff --git a/libavfilter/dnn/dnn_backend_common.h b/libavfilter/dnn/dnn_backend_common.h
index 9f5d37b3e0..3629cb4922 100644
--- a/libavfilter/dnn/dnn_backend_common.h
+++ b/libavfilter/dnn/dnn_backend_common.h
@@ -157,4 +157,13 @@ DNNAsyncStatusType ff_dnn_get_result_common(Queue *task_queue, AVFrame **in, AVF
  */
 int ff_dnn_fill_gettingoutput_task(TaskItem *task, DNNExecBaseParams *exec_params, void *backend_model, int input_height, int input_width, void *ctx);
 
+/**
+ * Check if the given frame contains a valid detection bounding box.
+ *
+ * @param frame The frame to check for valid detection bounding box
+ * 
+ * @return Non-zero if frame contains valid detection bounding box, 0 otherwise
+ */
+int ff_dnn_contain_valid_detection_bbox(AVFrame *frame);
+
 #endif
diff --git a/libavfilter/dnn/dnn_backend_openvino.c b/libavfilter/dnn/dnn_backend_openvino.c
index 2f6706dcd4..03183af822 100644
--- a/libavfilter/dnn/dnn_backend_openvino.c
+++ b/libavfilter/dnn/dnn_backend_openvino.c
@@ -1151,43 +1151,6 @@ static int get_input_ov(DNNModel *model, DNNData *input, const char *input_name)
 #endif
 }
 
-static int contain_valid_detection_bbox(AVFrame *frame)
-{
-    AVFrameSideData *sd;
-    const AVDetectionBBoxHeader *header;
-    const AVDetectionBBox *bbox;
-
-    sd = av_frame_get_side_data(frame, AV_FRAME_DATA_DETECTION_BBOXES);
-    if (!sd) { // this frame has nothing detected
-        return 0;
-    }
-
-    if (!sd->size) {
-        return 0;
-    }
-
-    header = (const AVDetectionBBoxHeader *)sd->data;
-    if (!header->nb_bboxes) {
-        return 0;
-    }
-
-    for (uint32_t i = 0; i < header->nb_bboxes; i++) {
-        bbox = av_get_detection_bbox(header, i);
-        if (bbox->x < 0 || bbox->w < 0 || bbox->x + bbox->w >= frame->width) {
-            return 0;
-        }
-        if (bbox->y < 0 || bbox->h < 0 || bbox->y + bbox->h >= frame->height) {
-            return 0;
-        }
-
-        if (bbox->classify_count == AV_NUM_DETECTION_BBOX_CLASSIFY) {
-            return 0;
-        }
-    }
-
-    return 1;
-}
-
 static int extract_lltask_from_task(DNNFunctionType func_type, TaskItem *task, Queue *lltask_queue, DNNExecBaseParams *exec_params)
 {
     switch (func_type) {
@@ -1217,7 +1180,7 @@ static int extract_lltask_from_task(DNNFunctionType func_type, TaskItem *task, Q
         task->inference_todo = 0;
         task->inference_done = 0;
 
-        if (!contain_valid_detection_bbox(frame)) {
+        if (!ff_dnn_contain_valid_detection_bbox(frame)) {
             return 0;
         }
 
-- 
2.34.1