Doni/OrangePi3588Media
5
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Make shoe detector box format configurable

Browse Source
This commit is contained in:
tian 2026-03-14 15:23:32 +08:00
parent 943228a244
commit 6d1456a9e1
2 changed files with 30 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
configs/sample_person_shoe_two_stage.json
View File

@ -74,19 +74,20 @@
"type": "ai_shoe_det",
"role": "filter",
"enable": true,
"infer_fps": 2,
"infer_fps": 1,
"model_path": "./models/shoe_detector_openimages_ppe_v1.rknn",
"model_w": 640,
"model_h": 640,
"conf": 0.25,
"nms": 0.45,
"v8_box_format": "cxcywh",
"append_detections": true,
"dynamic_roi": {
"enable": true,
"person_class_id": 0,
"shoe_class_id": 1,
"debug_roi_class_id": 2,
"max_rois": 4,
"max_rois": 3,
"min_person_height": 80,
"x_offset": -0.20,
"y_offset": 0.68,

30
plugins/ai_shoe_det/ai_shoe_det_node.cpp
View File

@ -283,6 +283,28 @@ public:
conf_thresh_ = config.ValueOr<float>("conf", 0.25f);
nms_thresh_ = config.ValueOr<float>("nms", 0.45f);
append_detections_ = config.ValueOr<bool>("append_detections", false);
{
const std::string bf = config.ValueOr<std::string>("v8_box_format", "auto");
if (bf == "xyxy") {
v8_box_format_ = V8BoxFormat::XyXy;
} else if (bf == "xywh") {
v8_box_format_ = V8BoxFormat::XyWh;
} else if (bf == "cxcywh") {
v8_box_format_ = V8BoxFormat::CxCyWh;
} else {
v8_box_format_ = V8BoxFormat::Auto;
}
}
{
const std::string act = config.ValueOr<std::string>("v8_cls_activation", "auto");
if (act == "sigmoid") {
v8_cls_activation_ = V8ClsActivation::Sigmoid;
} else if (act == "none") {
v8_cls_activation_ = V8ClsActivation::None;
} else {
v8_cls_activation_ = V8ClsActivation::Auto;
}
}
infer_interval_ms_ = std::max<int64_t>(
0, static_cast<int64_t>(config.ValueOr<int>("infer_interval_ms", 0)));
@ -581,7 +603,7 @@ private:
auto append_box = [&](float a, float b, float c, float d, float conf) {
if (!std::isfinite(a) || !std::isfinite(b) || !std::isfinite(c) || !std::isfinite(d)) return;
float x = 0.0f, y = 0.0f, w = 0.0f, h = 0.0f;
DecodeV8Box(a, b, c, d, model_w_, model_h_, V8BoxFormat::Auto, x, y, w, h);
DecodeV8Box(a, b, c, d, model_w_, model_h_, v8_box_format_, x, y, w, h);
if (!std::isfinite(x) || !std::isfinite(y) || !std::isfinite(w) || !std::isfinite(h)) return;
if (w <= 1e-3f || h <= 1e-3f) return;
if (w > model_w_ * 1.2f || h > model_h_ * 1.2f) return;
@ -599,7 +621,7 @@ private:
if (outputs[0].type == RKNN_TENSOR_FLOAT32) {
const float* data = reinterpret_cast<const float*>(outputs[0].data);
const bool apply_sigmoid = ResolveV8ApplySigmoid(
data, num_boxes, kNumClasses, layout.channels_first, V8ClsActivation::Auto);
data, num_boxes, kNumClasses, layout.channels_first, v8_cls_activation_);
for (int i = 0; i < num_boxes; ++i) {
const float score = layout.channels_first ? data[4 * num_boxes + i]
: data[i * num_channels + 4];
@ -618,7 +640,7 @@ private:
fp32_decode_buf_[i] = Fp16ToFp32(data[i]);
}
const bool apply_sigmoid = ResolveV8ApplySigmoid(
fp32_decode_buf_.data(), num_boxes, kNumClasses, layout.channels_first, V8ClsActivation::Auto);
fp32_decode_buf_.data(), num_boxes, kNumClasses, layout.channels_first, v8_cls_activation_);
for (int i = 0; i < num_boxes; ++i) {
const float score = layout.channels_first ? fp32_decode_buf_[4 * num_boxes + i]
: fp32_decode_buf_[i * num_channels + 4];
@ -733,6 +755,8 @@ private:
float conf_thresh_ = 0.25f;
float nms_thresh_ = 0.45f;
bool append_detections_ = false;
V8BoxFormat v8_box_format_ = V8BoxFormat::Auto;
V8ClsActivation v8_cls_activation_ = V8ClsActivation::Auto;
int64_t infer_interval_ms_ = 0;
int64_t last_infer_pts_ms_ = 0;
DynamicRoiConfig dynamic_roi_;