espressif / esp-dl

Espressif deep-learning library for AIoT applications
MIT License
519 stars 116 forks source link

Image_get_valid_boxes change (?) #16

Closed caleb221 closed 4 years ago

caleb221 commented 4 years ago

Hi! making an adaptation of your implementation for image pre/post processing for Pnet (I'm setting up MTCNN for a different purpose). I run into the following compilation error (fixed by commenting out config->threshold.score when i call get_valid_boxes. So, has there been any changes in the library that are different in the repo? Or am i missing something quite obvious?

error: too many arguments to function 'image_get_valid_boxes' origin_head[i] = image_get_valid_boxes(....)

Running the compiled form (as shown in the code below) results in a Guru Error

               Guru Meditation Error: Core  0 panic'ed (LoadProhibited). Exception was unhandled.
               Core 0 register dump:

-code image_resize_linear( .. ) ... printf("\n[PNET FWD] RESIZED\n"); in->h = height; in->w = width; in->stride = in->w * in->c; printf("\nSET UP IMAGE\n "); xTaskCreatePinnedToCore(&pnetCore1,"core2Print",50000,in,5,&pHandle,1); //vTaskDelete(&pHandle);

vTaskDelay(1000/portTICK_PERIOD_MS); printf("\n[PNET FWD] BACK FROM convolutions\n\n"); printf("\n\n..config w: %i \n",(&config->w)); // if (output ) // { printf("\n\nSKELETON KEY: %p\n\n",skeletonKey); //xSemaphoreTake(skeletonKey,portMAX_DELAY); printf("\n\n..getting valid boxes..\n\n\n");//%i.",(&config->w));

        origin_head[i] = image_get_valid_boxes(output->category->item,
                                               output->offset->item,
                                               NULL,
                                               output->category->w,
                                               output->category->h,
                                               1,
                                               &config->w,
                                              // config->threshold.score,
                                               2,
                                               scale,
                                               scale,
                                               false);

...and so on with further processing

--found the compilation error, but still getting a Guru load error compilation fix: origin_head[i] = image_get_valid_boxes( output->category->item, output->offset->item, NULL, output->category->w, output->category->h, 1, &config->w, config->threshold.score, 2, scale, //scale, false);

caleb221 commented 4 years ago

My mistake, I found the error