Luckfox pico Max Face Recognition

2024-03-26 8:16

rajadey150@gmail.com wrote: ↑2024-03-26 7:58 Sir,
I have followed your previous steps with required modifications to Cmakelists
Thus,
Please suggest

1.When to execute -killall rkipc(before compling the demo or after compling) or at the start?
2.while compling we should use "luckfox_retinaface_facenet_spi_demo_for_promax" instead here "./luckfox_retinaface_facenet_demo_test ./model/RetinaFace.rknn ./model/mobilefacenet.rknn ./test.jpg"
"luckfox_retinaface_facenet_spi_demo_for_promax ./model/RetinaFace.rknn ./model/mobilefacenet.rknn ./test.jpg

The command "killall rkipc" should be run before executing the demo, and it only needs to be executed once after the Luckfox device boots up.

The method of running the demo mentioned in the Readme.md is just using one of the demos (framebuffer mode) as an example. In reality, the command you should execute on the Luckfox-Pico is:

Code: Select all

cd luckfox_retinaface_facenet_spi_demo_for_promax #（Alternatively, you can use the luckfox_retinaface_facenet_spi_demo that you compiled yourself.）
./luckfox_retinaface_facenet_spi_demo_test ./model/RetinaFace.rknn ./model/mobilefacenet.rknn ./test.jpg

The test.jpg file needs to be provided by you.

2024-03-26 8:45

Respected Sir,
I have run your code with previous instructions
But no video or camera output is coming
Ubuntu output is this:-------------------------

# cd luckfox_retinaface_facenet_demo
# chmod a+x ./luckfox_retinaface_facenet_demo_test
# killall rkipc
# ./luckfox_retinaface_facenet_demo_test ./model/RetinaFace.rknn ./model/mobilef
acenet.rknn ./test.jpg
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
init retinaface
init facenet
E RKNN: failed to allocate fd, ret: -1, errno: 12, errstr: Cannot allocate memory
E RKNN: failed to allocate model memory!, size: 4426496, flags: #2
Retinaface Info
model input num: 1, output num: 3
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 640, 640, 3], n_elems=1228800, size=1228800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=1.000000
output tensors:
index=0, name=515, n_dims=3, dims=[1, 16800, 4, 0], n_elems=67200, size=67200, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-125, scale=1.951411
index=1, name=553, n_dims=3, dims=[1, 16800, 2, 0], n_elems=33600, size=33600, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-2, scale=0.243586
index=2, name=592, n_dims=3, dims=[1, 16800, 10, 0], n_elems=168000, size=168000, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-119, scale=1.485194
input_attrs[0].size_with_stride=1228800
output mem [0] = 67200
output mem [1] = 33600
output mem [2] = 168000
model is NHWC input fmt
Facenet Info
model input num: 0, output num: 0
input tensors:
output tensors:
input_attrs[0].size_with_stride=67200
input_mems rknn_set_io_mem fail! ret=-7

2024-03-26 8:50

after few modifications this output is coming
But Still no camera or video interference is coming

# chmod a+x ./luckfox_retinaface_facenet_demo_test
# killall rkipc
# ./luckfox_retinaface_facenet_demo_test ./model/RetinaFace.rknn ./model/mobilef
acenet.rknn ./test.jpg
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
init retinaface
init facenet
Retinaface Info
model input num: 1, output num: 3
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 640, 640, 3], n_elems=1228800, size=1228800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=1.000000
output tensors:
index=0, name=515, n_dims=3, dims=[1, 16800, 4, 0], n_elems=67200, size=67200, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-125, scale=1.951411
index=1, name=553, n_dims=3, dims=[1, 16800, 2, 0], n_elems=33600, size=33600, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-2, scale=0.243586
index=2, name=592, n_dims=3, dims=[1, 16800, 10, 0], n_elems=168000, size=168000, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-119, scale=1.485194
input_attrs[0].size_with_stride=1228800
output mem [0] = 67200
output mem [1] = 33600
output mem [2] = 168000
model is NHWC input fmt
Facenet Info
model input num: 1, output num: 1
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 160, 160, 3], n_elems=76800, size=76800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003922
output tensors:
index=0, name=313, n_dims=2, dims=[1, 128, 0, 0], n_elems=128, size=128, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=9, scale=0.081335
input_attrs[0].size_with_stride=76800
output mem [0] = 128
Init success

2024-03-26 9:11

rajadey150@gmail.com wrote: ↑2024-03-26 8:50 after few modifications this output is coming
But Still no camera or video interference is coming

# chmod a+x ./luckfox_retinaface_facenet_demo_test
# killall rkipc
# ./luckfox_retinaface_facenet_demo_test ./model/RetinaFace.rknn ./model/mobilef
acenet.rknn ./test.jpg
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
init retinaface
init facenet
Retinaface Info
model input num: 1, output num: 3
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 640, 640, 3], n_elems=1228800, size=1228800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=1.000000
output tensors:
index=0, name=515, n_dims=3, dims=[1, 16800, 4, 0], n_elems=67200, size=67200, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-125, scale=1.951411
index=1, name=553, n_dims=3, dims=[1, 16800, 2, 0], n_elems=33600, size=33600, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-2, scale=0.243586
index=2, name=592, n_dims=3, dims=[1, 16800, 10, 0], n_elems=168000, size=168000, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-119, scale=1.485194
input_attrs[0].size_with_stride=1228800
output mem [0] = 67200
output mem [1] = 33600
output mem [2] = 168000
model is NHWC input fmt
Facenet Info
model input num: 1, output num: 1
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 160, 160, 3], n_elems=76800, size=76800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003922
output tensors:
index=0, name=313, n_dims=2, dims=[1, 128, 0, 0], n_elems=128, size=128, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=9, scale=0.081335
input_attrs[0].size_with_stride=76800
output mem [0] = 128
Init success

Congratulations on reaching this stage, you're close to success! You've successfully loaded the RKNN model.
However, I've noticed that you're using a program adapted for the framebuffer instead of one adapted for Restouch. This program fails to run because it can't find the framebuffer device during initialization.

A properly adapted program for Restouch will output the type of development board.

Please use the correct demo for validation.

Code: Select all

 ./luckfox_retinaface_facenet_spi_demo_test ./model/RetinaFace.rknn ./model/mobilef
acenet.rknn ./test.jpg

This is the correct demo program.

: luckfox_retinaface_facenet_spi_demo_for_promax.7z (4.24 MiB)
Downloaded 525 times

2024-03-26 9:20

Thank you sir help
Still few doubts to be cleared.
1.Do i need to modify the CMakelist?
2.Do i need to give the path name in CMake list?
3.Will i compile this code folder same as previous one?

2024-03-26 9:48

Sir,
I have tried your last instructions,

But this output is coming

# ./luckfox_retinaface_facenet_spi_demo_test ./model/RetinaFace.rknn ./model/mob
ilefacenet.rknn .test.jpg
/bin/sh: ./luckfox_retinaface_facenet_spi_demo_test: not found
#

2024-03-26 9:51

rajadey150@gmail.com wrote: ↑2024-03-26 9:20 Thank you sir help
Still few doubts to be cleared.
1.Do i need to modify the CMakelist?
2.Do i need to give the path name in CMake list?
3.Will i compile this code folder same as previous one?

I'm very concerned about whether you can run the program normally now. Can you tell me if you can run it normally?

1.If you need to use Luckfox-Pico Pro/Max, you'll need to modify the CMakeLists.txt file because I've enabled support for Luckfox-Pico Plus by default. You just need to choose whether to keep the line add_definitions(-DLUCKFOX_PICO_PLUS).

2.I've already specified the location of the output folder in the CMakeLists.txt file. After compilation, you can find the output folder luckfox_retinaface_facenet_spi_demo in the parent directory (example/luckfox_retinaface_facenet/cpp_spi).

3。I'm not sure which folder you're referring to. The attachment in my previous response contains the compiled output files, and the compilation work was all done in luckfox-pico-rknn-example/example.

2024-03-26 9:56

# ./luckfox_retinaface_facenet_spi_demo_test ./model/RetinaFace.rknn ./model/mob
ilefacenet.rknn ./test.jpg
/bin/sh: ./luckfox_retinaface_facenet_spi_demo_test: not found
Sir as you see,after altering the compile version this error is showing

Now, if i run this............no Camera or LCD interface is on
chmod a+x ./luckfox_retinaface_facenet_demo_test
# ./luckfox_retinaface_facenet_demo_test ./model/RetinaFace.rknn ./model/mobilef
acenet.rknn ./test.jpg
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for jpeg decoder cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
this device is not whitelisted for capture cvi
init retinaface
init facenet
Retinaface Info
model input num: 1, output num: 3
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 640, 640, 3], n_elems=1228800, size=1228800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=1.000000
output tensors:
index=0, name=515, n_dims=3, dims=[1, 16800, 4, 0], n_elems=67200, size=67200, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-125, scale=1.951411
index=1, name=553, n_dims=3, dims=[1, 16800, 2, 0], n_elems=33600, size=33600, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-2, scale=0.243586
index=2, name=592, n_dims=3, dims=[1, 16800, 10, 0], n_elems=168000, size=168000, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=-119, scale=1.485194
input_attrs[0].size_with_stride=1228800
output mem [0] = 67200
output mem [1] = 33600
output mem [2] = 168000
model is NHWC input fmt
Facenet Info
model input num: 1, output num: 1
input tensors:
index=0, name=input.1, n_dims=4, dims=[1, 160, 160, 3], n_elems=76800, size=76800, fmt=NHWC, type=INT8, qnt_type=AFFINE, zp=-128, scale=0.003922
output tensors:
index=0, name=313, n_dims=2, dims=[1, 128, 0, 0], n_elems=128, size=128, fmt=UNDEFINED, type=INT8, qnt_type=AFFINE, zp=9, scale=0.081335
input_attrs[0].size_with_stride=76800
output mem [0] = 128
Init success
Which is to modify for SPI version for 3.5LCD restouch

2024-03-26 9:56

rajadey150@gmail.com wrote: ↑2024-03-26 9:48 Sir,
I have tried your last instructions,

But this output is coming

# ./luckfox_retinaface_facenet_spi_demo_test ./model/RetinaFace.rknn ./model/mob
ilefacenet.rknn .test.jpg
/bin/sh: ./luckfox_retinaface_facenet_spi_demo_test: not found
#

Did you use this demo I provided?

2024-03-26 10:06

Yes sir,
I have downloaded the code,

Here is my doubt ,

1.Do i have adb push luckfox_retinaface_facenet_spi_demo_for_promax to my DevBoard or
adb push luckfox_retinaface_facenet_demo?

if so then confirm?