MLECO-2188: Review documentation
Change-Id: I15dceeafec554e58257e8e6d6e8896cc9b397106
diff --git a/docs/use_cases/ad.md b/docs/use_cases/ad.md
index babbb93..f953f72 100644
--- a/docs/use_cases/ad.md
+++ b/docs/use_cases/ad.md
@@ -311,13 +311,13 @@
User input required
Enter option number from:
-1. Classify next audio clip
-2. Classify audio clip at chosen index
-3. Run classification on all audio clips
-4. Show NN model info
-5. List audio clips
+ 1. Classify next audio signal
+ 2. Classify audio signal at chosen index
+ 3. Run classification on all audio signals
+ 4. Show NN model info
+ 5. List audio signals
-Choice:
+ Choice:
```
@@ -328,7 +328,7 @@
2. Classify audio clip at chosen index: Runs inference on the chosen audio clip.
> **Note:** Please make sure to select audio clip index within the range of supplied audio clips during application
- > build. By default, a pre-built application has four files, with indexes from `0` to `3`.
+ > build. By default, the pre-built application has one file with index `0`.
3. Run ... on all: Triggers sequential inference executions on all built-in applications.
@@ -354,7 +354,7 @@
INFO - Quant dimension: 0
INFO - Scale[0] = 0.048891
INFO - ZeroPoint[0] = -30
- INFO - Activation buffer (a.k.a tensor arena) size used: 198016
+ INFO - Activation buffer (a.k.a tensor arena) size used: 321252
INFO - Number of operators: 1
INFO - Operator 0: ethos-u
```
@@ -363,10 +363,7 @@
```log
INFO - List of Files:
- INFO - 0 =>; anomaly_id_00_00000000.wav
- INFO - 1 =>; anomaly_id_02_00000076.wav
- INFO - 2 =>; normal_id_00_00000004.wav
- INFO - 3 =>; normal_id_02_00000001.wav
+ INFO - 0 =>; random_id_00_000000.wav
```
### Running Anomaly Detection
@@ -376,53 +373,41 @@
The following example illustrates the output of an application:
```log
-INFO - Running inference on audio clip 0 => anomaly_id_00_00000000.wav
-INFO - Inference 1/13
-INFO - Inference 2/13
-INFO - Inference 3/13
-INFO - Inference 4/13
-INFO - Inference 5/13
-INFO - Inference 6/13
-INFO - Inference 7/13
-INFO - Inference 8/13
-INFO - Inference 9/13
-INFO - Inference 10/13
-INFO - Inference 11/13
-INFO - Inference 12/13
-INFO - Inference 13/13
-INFO - Average anomaly score is: -0.024493
-Anomaly threshold is: -0.800000
-Anomaly detected!
+INFO - Running inference on audio clip 4 => random_id_00_000000.wav
+INFO - Inference 1/1
+INFO - Average anomaly score is: -0.883147
+INFO - Anomaly threshold is: -0.800000
+INFO - Everything fine, no anomaly detected!
INFO - Profile for Inference:
-INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 628122
-INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 135087
-INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 62870
-INFO - NPU ACTIVE cycles: 1081007
-INFO - NPU IDLE cycles: 626
-INFO - NPU TOTAL cycles: 1081634
+INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 485194
+INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 138111
+INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 56245
+INFO - NPU ACTIVE cycles: 967731
+INFO - NPU IDLE cycles: 441
+INFO - NPU TOTAL cycles: 968172
```
As multiple inferences must be run for one clip, it takes around a minute for all inferences to complete.
-For the `anomaly_id_00_00000000.wav` clip, after averaging results across all inferences, the score is greater than the
-chosen anomaly threshold. Therefore, an anomaly was detected with the machine in this clip.
+For the `random_id_00_000000.wav` clip, after averaging results across all inferences needed, the score is less than the
+chosen anomaly threshold. Therefore, an anomaly was not detected with the machine in this clip.
The profiling section of the log shows that for each inference. For the last inference, the profiling reports:
- *Ethos-U* PMU report:
- - 1,081,634 total cycle: The number of NPU cycles
+ - 968,172 total cycle: The number of NPU cycles
- - 1,081,007 active cycles: number of NPU cycles that were used for computation
+ - 967,731 active cycles: number of NPU cycles that were used for computation
- - 626 idle cycles: number of cycles for which the NPU was idle
+ - 441 idle cycles: number of cycles for which the NPU was idle
- - 628,122 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where
+ - 48,5194 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where
*Ethos-U* NPU reads and writes to the computation buffers, activation buf, or tensor arenas.
- - 135,087 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus.
+ - 138,111 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus.
- - 62,870 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where
+ - 56,245 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where
*Ethos-U* NPU reads the model. So, read-only.
- For FPGA platforms, a CPU cycle count can also be enabled. However, do not use cycle counters for FVP, as the CPU
diff --git a/docs/use_cases/img_class.md b/docs/use_cases/img_class.md
index 13856e8..ea70ea5 100644
--- a/docs/use_cases/img_class.md
+++ b/docs/use_cases/img_class.md
@@ -295,11 +295,11 @@
User input required
Enter option number from:
-1. Classify next image
-2. Classify image at chosen index
-3. Run classification on all images
-4. Show NN model info
-5. List images
+ 1. Classify next ifm
+ 2. Classify ifm at chosen index
+ 3. Run classification on all ifm
+ 4. Show NN model info
+ 5. List ifm
Choice:
diff --git a/docs/use_cases/visual_wake_word.md b/docs/use_cases/visual_wake_word.md
index 299a8ff..a011d5e 100644
--- a/docs/use_cases/visual_wake_word.md
+++ b/docs/use_cases/visual_wake_word.md
@@ -16,6 +16,7 @@
## Introduction
This document describes the process of setting up and running the Arm® Ethos™-U NPU Visual Wake Word example.
+Visual Wake Words is a common vision use-case to detect if a the provided image contains a person.
Use case code could be found in [source/use_case/vww](../../source/use_case/vww) directory.
@@ -57,7 +58,7 @@
### Build process
> **Note:** This section describes the process for configuring the build for `MPS3: SSE-300` for different target
->platform see [Building](../documentation.md#Building) section.
+> platform see [Building](../documentation.md#Building) section.
Create a build directory and navigate inside:
@@ -274,11 +275,11 @@
User input required
Enter option number from:
-1. Classify next image
-2. Classify image at chosen index
-3. Run classification on all images
-4. Show NN model info
-5. List images
+ 1. Classify next ifm
+ 2. Classify ifm at chosen index
+ 3. Run classification on all ifm
+ 4. Show NN model info
+ 5. List ifm
Choice:
@@ -297,57 +298,56 @@
4. “Show NN model info” menu option prints information about model data type, input and output tensor sizes:
```log
- INFO - uTFL version: 2.5.0
- INFO - Added ethos-u support to op resolver
- INFO - Creating allocator using tensor arena in SRAM
- INFO - Allocating tensors
- INFO - Model INPUT tensors:
- INFO - tensor type is INT8
- INFO - tensor occupies 16384 bytes with dimensions
- INFO - 0: 1
- INFO - 1: 128
- INFO - 2: 128
- INFO - 3: 1
- INFO - Quant dimension: 0
- INFO - Scale[0] = 0.008138
- INFO - ZeroPoint[0] = -70
- INFO - Model OUTPUT tensors:
- INFO - tensor type is INT8
- INFO - tensor occupies 2 bytes with dimensions
- INFO - 0: 1
- INFO - 1: 2
- INFO - Quant dimension: 0
- INFO - Scale[0] = 0.022299
- INFO - ZeroPoint[0] = -17
- INFO - Activation buffer (a.k.a tensor arena) size used: 133292
- INFO - Number of operators: 19
- INFO - Operator 0: ethos-u
- INFO - Operator 1: PAD
- INFO - Operator 2: ethos-u
- INFO - Operator 3: PAD
- INFO - Operator 4: ethos-u
- INFO - Operator 5: PAD
- INFO - Operator 6: ethos-u
- INFO - Operator 7: PAD
- INFO - Operator 8: ethos-u
- INFO - Operator 9: PAD
- INFO - Operator 10: ethos-u
- INFO - Operator 11: PAD
- INFO - Operator 12: ethos-u
- INFO - Operator 13: PAD
- INFO - Operator 14: ethos-u
- INFO - Operator 15: PAD
- INFO - Operator 16: ethos-u
- INFO - Operator 17: AVERAGE_POOL_2D
- INFO - Operator 18: ethos-u
+ INFO - Added ethos-u support to op resolver
+ INFO - Creating allocator using tensor arena in SRAM
+ INFO - Allocating tensors
+ INFO - Model INPUT tensors:
+ INFO - tensor type is INT8
+ INFO - tensor occupies 16384 bytes with dimensions
+ INFO - 0: 1
+ INFO - 1: 128
+ INFO - 2: 128
+ INFO - 3: 1
+ INFO - Quant dimension: 0
+ INFO - Scale[0] = 0.008138
+ INFO - ZeroPoint[0] = -70
+ INFO - Model OUTPUT tensors:
+ INFO - tensor type is INT8
+ INFO - tensor occupies 2 bytes with dimensions
+ INFO - 0: 1
+ INFO - 1: 2
+ INFO - Quant dimension: 0
+ INFO - Scale[0] = 0.022299
+ INFO - ZeroPoint[0] = -17
+ INFO - Activation buffer (a.k.a tensor arena) size used: 133716
+ INFO - Number of operators: 19
+ INFO - Operator 0: ethos-u
+ INFO - Operator 1: PAD
+ INFO - Operator 2: ethos-u
+ INFO - Operator 3: PAD
+ INFO - Operator 4: ethos-u
+ INFO - Operator 5: PAD
+ INFO - Operator 6: ethos-u
+ INFO - Operator 7: PAD
+ INFO - Operator 8: ethos-u
+ INFO - Operator 9: PAD
+ INFO - Operator 10: ethos-u
+ INFO - Operator 11: PAD
+ INFO - Operator 12: ethos-u
+ INFO - Operator 13: PAD
+ INFO - Operator 14: ethos-u
+ INFO - Operator 15: PAD
+ INFO - Operator 16: ethos-u
+ INFO - Operator 17: AVERAGE_POOL_2D
+ INFO - Operator 18: ethos-u
```
5. “List Images” menu option prints a list of pair image indexes - the original filenames embedded in the application:
```log
- INFO - List of images:
- INFO - 0 => adult_blur.png
- INFO - 1 => man_in_red_jacket.png
+ INFO - List of Files:
+ INFO - 0 => man_in_red_jacket.png
+ INFO - 1 => st_paul_s_cathedral.png
```
### Running Visual Wake Word
@@ -357,40 +357,40 @@
The following example illustrates application output for classification:
```log
-INFO - Running inference on image 0 => adult_blur.png
+INFO - Running inference on image 0 => man_in_red_jacket.png
INFO - Final results:
-INFO -Total number of inferences: 1
-INFO - 0) 0 (0.601562) -> Person detected: No
-INFO - Profile for Inference :
-INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 95992
-INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 59735
-INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 34477
-INFO - NPU ACTIVE cycles: 372782
-INFO - NPU IDLE cycles: 390
-INFO - NPU total cycles: 373172
+INFO - Total number of inferences: 1
+INFO - 0) 1 (3.211100) -> person
+INFO - Profile for Inference:
+INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 228679
+INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 153031
+INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 40625
+INFO - NPU ACTIVE cycles: 706754
+INFO - NPU IDLE cycles: 10954
+INFO - NPU TOTAL cycles: 717708
```
It could take several minutes to complete one inference run (average time is 2-3 minutes).
-The log shows the inference results for “image 1” (1 - index) that corresponds to “adult_blur.png” in the sample
+The log shows the inference results for “image 1” (1 - index) that corresponds to man_in_red_jacket.png” in the sample
image resource folder.
The profiling section of the log shows that for this inference:
- Ethos-U's PMU report:
- - 373,172 total cycle: The number of NPU cycles
+ - 717,708 total cycle: The number of NPU cycles
- - 372,782 active cycles: number of NPU cycles that were used for computation
+ - 706,754 active cycles: number of NPU cycles that were used for computation
- - 390 idle cycles: number of cycles for which the NPU was idle
+ - 10,954 idle cycles: number of cycles for which the NPU was idle
- - 95,992 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where
+ - 228,679 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where
Ethos-U NPU reads and writes to the computation buffers (activation buf/tensor arenas).
- - 59,735 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus.
+ - 153,031 AXI0 write beats: The number of AXI beats with write transactions to AXI0 bus.
- - 34,477 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where
+ - 40,625 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where
Ethos-U NPU reads the model (read only)
- For FPGA platforms, CPU cycle count can also be enabled. For FVP, however, CPU cycle counters should not be used as
diff --git a/release_notes.txt b/release_notes.txt
index da3a5e1..5bd8137 100644
--- a/release_notes.txt
+++ b/release_notes.txt
@@ -1,7 +1,9 @@
Changes in 21.11
* Support for 21.08 Ethos-U component and dependencies (core-software, core-driver, Vela 3.1.0, CMSIS, TensorFlow Lite).
* Added dynamic load support for FVP for inference runner use-case.
- * Added support for different memory modes: Shared_Sram, Dedicated_Sram and Sram_Only
+ * Added support for different memory modes: Shared_Sram, Dedicated_Sram and Sram_Only.
+ * Added Noise Reduction use case.
+ * Updated support for Arm GNU Embedded Toolchain 10.3-2021.07 and Arm Compiler 6.17.
* Documentation updates.
Changes in 21.08