Deploy-SSD-ON-STM32

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

Deploy-SSD-ON-STM32

(★ 39)

An interesting project. Deploying SSD object detection model on STM32H7.

Deploy-SSD-ON-STM32 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Deploy-SSD-ON-STM32

*An interesting project. Deploying SSD Object Detection Models on the STM32*

**Introduction video : [(开源+教程)在STM32上实时推理SSD目标检测网络](https://www.bilibili.com/video/BV1mxmaYwEXb/?spm_id_from=333.999.0.0&vd_source=efd9304d68124acc06902219a989f4fb)**

**Looking forward to your support!**

### 👑 Describption

#### 🎉Model

- st_ssd_mobilenet_v1    ( .h5  ---->  .tflite  ---->  STM32CubeAI quantized)
- Train on keras, use Adam(lr=0.001), 500 epochs.
- Input_shape : (192, 192, 3) (RGB888 data)
- Output_shape (It has three output) :
    1. cls : (1, 3830, 2) 
    2. boxes : (1, 3830, 4)
    3. anchors : (1, 3830, 4)
- Input quantization : uint8 
- Output quantization: Float

#### 🥇MCU

- STM32H750XBH6 (STM32H743 also can run, if the memory enough) **480MHz**
- ROM : 128 kB (Actually, it has 2 mB, as same as STM32H743XI)
- RAM : 1 mB 
- Generated by STM32CubeMX and STM32CubeAI

#### 🎨Project

- Used Flash : 580.31 kB
- Used RAM   : 607.76 kB
- Infer Cost (Excluding postprocess): 190 ms (Test by STM32H723ZG **550MHz**)

### 📂 Folder Structure

- deployment on mcu: Contains code and files for deploying the SSD model on STM32.

- model: Includes model files for training and conversion.

- train: Stores scripts for model training using Keras.

### 🧐 Acknowledgment

Many thanks to [STM32 AI Model Zoo](https://github.com/STMicroelectronics/stm32ai-modelzoo) by STMicroelectronics for their valuable work!

Deploy-SSD-ON-STM32

Deploy-SSD-ON-STM32 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Deploy-SSD-ON-STM32

Deploy-SSD-ON-STM32 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README