JuanJoseMoralesCalvo

To create a BLE input stream and output stream i recommend this tutorial (kotlin) : https://www.youtube.com/watch?v=Oz4CBHrxMMs

The model was trained with a batch of 64 images just if you need more info about it. @impjdi

Same error with those changes: org.tensorflow.lite.examples.objectdetection E/tflite: Following operations are not supported by GPU delegate: CUSTOM TFLite_Detection_PostProcess: TFLite_Detection_PostProcess PACK: OP is supported, but tensor type/shape isn't compatible. RESHAPE: OP is...

Yes, here it is the tflite model with the metadata needed: [detectlast.zip](https://github.com/tensorflow/tensorflow/files/9109516/detectlast.zip)

Im a bit knew with these :D, could you guide me a little bit with the solutions? Im sorry for taking your time.

The workaround for me was to add security to the BLE, but honestly i dont know the reason :D Attach to setup() after advertising: NimBLESecurity *SECURITY_MANDATORY = new BLESecurity(); SECURITY_MANDATORY->setAuthenticationMode(ESP_LE_AUTH_BOND);

@CybershoesVR Try this workaround from chewara, it sounds like its happening the same: class MyCallbacks : public BLEServerCallbacks { void onConnect(BLEServer* pServer){ Serial.println("connected"); BLEDescriptor *desc = input->getDescriptorByUUID(BLEUUID((uint16_t)0x2902)); uint8_t val[] =...

You have to add something like this on the onConenct: BLEDescriptor *desc = input->getDescriptorByUUID(BLEUUID((uint16_t)0x2902)); uint8_t val[] = {0x01, 0x00}; desc->setValue(val, 2);

The workaround on comment #12 didnt work as i understand its only for the nkolban esp32 libraries. Finally by adding security it worked. NimBLESecurity *SECURITY_MANDATORY = new NimBLESecurity(); SECURITY_MANDATORY->setAuthenticationMode(ESP_LE_AUTH_BOND); //...

@chegewara I know that im just answering myself but if anyone need this i made the workaround like this: void onConnect(BLEServer* BLEservidor){ // Respuesta a error de post-emparejamiento ble if(digitalRead(17)...