virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-28 20:53:54 +02:00
Code Issues Releases Activity

libavfilter/vf_dnn_detect: Add yolov4 support

Browse Source 
The difference of yolov4 is that sigmoid function needed to be applied
on x, y coordinates. Also make it compatiple with NHWC output as the
yolov4 model from openvino model zoo has NHWC output layout.

Model refer to: https://github.com/openvinotoolkit/open_model_zoo/tree/master/models/public/yolo-v4-tf

Signed-off-by: Wenbin Chen <wenbin.chen@intel.com>
Reviewed-by: Guo Yejun <yejun.guo@intel.com>
This commit is contained in:
Wenbin Chen 2023-12-12 10:33:34 +08:00 committed by Guo Yejun
parent a882fc0294
commit 1fa3346c70
1 changed files with 59 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

71
libavfilter/vf_dnn_detect.c
View File

@ -35,7 +35,8 @@
typedef enum { typedef enum {
DDMT_SSD, DDMT_SSD,
DDMT_YOLOV1V2, DDMT_YOLOV1V2,
DDMT_YOLOV3 DDMT_YOLOV3,
DDMT_YOLOV4
} DNNDetectionModelType; } DNNDetectionModelType;
typedef struct DnnDetectContext { typedef struct DnnDetectContext {
@ -75,6 +76,7 @@ static const AVOption dnn_detect_options[] = {
{ "ssd", "output shape [1, 1, N, 7]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_SSD }, 0, 0, FLAGS, "model_type" }, { "ssd", "output shape [1, 1, N, 7]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_SSD }, 0, 0, FLAGS, "model_type" },
{ "yolo", "output shape [1, N*Cx*Cy*DetectionBox]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_YOLOV1V2 }, 0, 0, FLAGS, "model_type" }, { "yolo", "output shape [1, N*Cx*Cy*DetectionBox]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_YOLOV1V2 }, 0, 0, FLAGS, "model_type" },
{ "yolov3", "outputs shape [1, N*D, Cx, Cy]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_YOLOV3 }, 0, 0, FLAGS, "model_type" }, { "yolov3", "outputs shape [1, N*D, Cx, Cy]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_YOLOV3 }, 0, 0, FLAGS, "model_type" },
{ "yolov4", "outputs shape [1, N*D, Cx, Cy]", 0, AV_OPT_TYPE_CONST, { .i64 = DDMT_YOLOV4 }, 0, 0, FLAGS, "model_type" },
{ "cell_w", "cell width", OFFSET2(cell_w), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS }, { "cell_w", "cell width", OFFSET2(cell_w), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS },
{ "cell_h", "cell height", OFFSET2(cell_h), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS }, { "cell_h", "cell height", OFFSET2(cell_h), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS },
{ "nb_classes", "The number of class", OFFSET2(nb_classes), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS }, { "nb_classes", "The number of class", OFFSET2(nb_classes), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INTMAX_MAX, FLAGS },
@ -84,6 +86,14 @@ static const AVOption dnn_detect_options[] = {
AVFILTER_DEFINE_CLASS(dnn_detect); AVFILTER_DEFINE_CLASS(dnn_detect);
static inline float sigmoid(float x) {
return 1.f / (1.f + exp(-x));
}
static inline float linear(float x) {
return x;
}
static int dnn_detect_get_label_id(int nb_classes, int cell_size, float *label_data) static int dnn_detect_get_label_id(int nb_classes, int cell_size, float *label_data)
{ {
float max_prob = 0; float max_prob = 0;
@ -147,6 +157,8 @@ static int dnn_detect_parse_yolo_output(AVFrame *frame, DNNData *output, int out
float *output_data = output[output_index].data; float *output_data = output[output_index].data;
float *anchors = ctx->anchors; float *anchors = ctx->anchors;
AVDetectionBBox *bbox; AVDetectionBBox *bbox;
float (*post_process_raw_data)(float x);
int is_NHWC = 0;
if (ctx->model_type == DDMT_YOLOV1V2) { if (ctx->model_type == DDMT_YOLOV1V2) {
cell_w = ctx->cell_w; cell_w = ctx->cell_w;
@ -154,13 +166,30 @@ static int dnn_detect_parse_yolo_output(AVFrame *frame, DNNData *output, int out
scale_w = cell_w; scale_w = cell_w;
scale_h = cell_h; scale_h = cell_h;
} else { } else {
cell_w = output[output_index].width; if (output[output_index].height != output[output_index].width &&
cell_h = output[output_index].height; output[output_index].height == output[output_index].channels) {
is_NHWC = 1;
cell_w = output[output_index].height;
cell_h = output[output_index].channels;
} else {
cell_w = output[output_index].width;
cell_h = output[output_index].height;
}
scale_w = ctx->scale_width; scale_w = ctx->scale_width;
scale_h = ctx->scale_height; scale_h = ctx->scale_height;
} }
box_size = nb_classes + 5; box_size = nb_classes + 5;
switch (ctx->model_type) {
case DDMT_YOLOV1V2:
case DDMT_YOLOV3:
post_process_raw_data = linear;
break;
case DDMT_YOLOV4:
post_process_raw_data = sigmoid;
break;
}
if (!cell_h || !cell_w) { if (!cell_h || !cell_w) {
av_log(filter_ctx, AV_LOG_ERROR, "cell_w and cell_h are detected\n"); av_log(filter_ctx, AV_LOG_ERROR, "cell_w and cell_h are detected\n");
return AVERROR(EINVAL); return AVERROR(EINVAL);
@ -198,19 +227,36 @@ static int dnn_detect_parse_yolo_output(AVFrame *frame, DNNData *output, int out
float *detection_boxes_data; float *detection_boxes_data;
int label_id; int label_id;
detection_boxes_data = output_data + box_id * box_size * cell_w * cell_h; if (is_NHWC) {
conf = detection_boxes_data[cy * cell_w + cx + 4 * cell_w * cell_h]; detection_boxes_data = output_data +
((cy * cell_w + cx) * detection_boxes + box_id) * box_size;
conf = post_process_raw_data(detection_boxes_data[4]);
} else {
detection_boxes_data = output_data + box_id * box_size * cell_w * cell_h;
conf = post_process_raw_data(
detection_boxes_data[cy * cell_w + cx + 4 * cell_w * cell_h]);
}
if (conf < conf_threshold) { if (conf < conf_threshold) {
continue; continue;
} }
x = detection_boxes_data[cy * cell_w + cx]; if (is_NHWC) {
y = detection_boxes_data[cy * cell_w + cx + cell_w * cell_h]; x = post_process_raw_data(detection_boxes_data[0]);
w = detection_boxes_data[cy * cell_w + cx + 2 * cell_w * cell_h]; y = post_process_raw_data(detection_boxes_data[1]);
h = detection_boxes_data[cy * cell_w + cx + 3 * cell_w * cell_h]; w = detection_boxes_data[2];
label_id = dnn_detect_get_label_id(ctx->nb_classes, cell_w * cell_h, h = detection_boxes_data[3];
detection_boxes_data + cy * cell_w + cx + 5 * cell_w * cell_h); label_id = dnn_detect_get_label_id(ctx->nb_classes, 1, detection_boxes_data + 5);
conf = conf * detection_boxes_data[cy * cell_w + cx + (label_id + 5) * cell_w * cell_h]; conf = conf * post_process_raw_data(detection_boxes_data[label_id + 5]);
} else {
x = post_process_raw_data(detection_boxes_data[cy * cell_w + cx]);
y = post_process_raw_data(detection_boxes_data[cy * cell_w + cx + cell_w * cell_h]);
w = detection_boxes_data[cy * cell_w + cx + 2 * cell_w * cell_h];
h = detection_boxes_data[cy * cell_w + cx + 3 * cell_w * cell_h];
label_id = dnn_detect_get_label_id(ctx->nb_classes, cell_w * cell_h,
detection_boxes_data + cy * cell_w + cx + 5 * cell_w * cell_h);
conf = conf * post_process_raw_data(
detection_boxes_data[cy * cell_w + cx + (label_id + 5) * cell_w * cell_h]);
}
bbox = av_mallocz(sizeof(*bbox)); bbox = av_mallocz(sizeof(*bbox));
if (!bbox) if (!bbox)
@ -410,6 +456,7 @@ static int dnn_detect_post_proc_ov(AVFrame *frame, DNNData *output, int nb_outpu
if (ret < 0) if (ret < 0)
return ret; return ret;
case DDMT_YOLOV3: case DDMT_YOLOV3:
case DDMT_YOLOV4:
ret = dnn_detect_post_proc_yolov3(frame, output, filter_ctx, nb_outputs); ret = dnn_detect_post_proc_yolov3(frame, output, filter_ctx, nb_outputs);
if (ret < 0) if (ret < 0)
return ret; return ret;