inviolus commited on
Commit
86b565f
·
verified ·
1 Parent(s): 4f7f2cc

Point to the non-commercial VoxHOD-NC branch

Browse files
Files changed (1) hide show
  1. README.md +9 -11
README.md CHANGED
@@ -14,18 +14,16 @@ tags:
14
  - centernet
15
  ---
16
 
17
- # VoxHOD: tiny aerial person detector
18
 
19
  A CenterNet-style, anchor-free person detector for downward-looking drone
20
  cameras: **136,547 parameters, 197 KB full-int8 TFLite**, or **37,363 parameters,
21
  75 KB** for the nano variant, built for ESP32-S3-class microcontrollers. It marks where people are in a 160×160 RGB
22
- tile so a drone can flag places worth a closer look. Search and rescue is the use it
23
- has been trained and measured for so far.
24
 
25
- > **Türkçe özet:** Havadan çekilen görüntüler için, mikrodenetleyicide
26
- > çalışacak kadar küçük (136 bin parametre, int8 dosyası 197 KB), genel amaçlı
27
- > bir insan tespit modeli. Şimdiye kadar arama kurtarma verileriyle eğitilip
28
- > ölçüldü. HERIDAL test setinde int8 hâli, eşik 0,2'de insanların
29
  > %83,9'unu buluyor. 75 KB'lık nano sürümü %81,2'sini buluyor. Eğitimde
30
  > kullanılmayan bir WiSARD klibinde bu oranlar %61 ve %62'ye düşüyor. CC BY 4.0.
31
 
@@ -109,6 +107,8 @@ people), never trained on. Treat it as a first signal.
109
 
110
  ## Data and license
111
 
 
 
112
  Trained on [HERIDAL](https://zenodo.org/records/5662351) (CC BY 3.0):
113
  D. Božić-Štulić, Ž. Marušić, S. Gotovac, "Deep Learning Approach in Aerial
114
  Imagery for Supporting Land Search and Rescue Missions", IJCV 2019.
@@ -121,7 +121,5 @@ Weights: **CC BY 4.0**. Training code: MIT.
121
 
122
  ## Intended use
123
 
124
- VoxHOD is a general-purpose aerial person detector. Search and rescue is the
125
- use it has been trained and measured for so far: helping find people who are
126
- lost or injured. Other settings are unmeasured. It is not intended for
127
- targeting or harming people.
 
14
  - centernet
15
  ---
16
 
17
+ # VoxHOD: tiny person detector for drone search and rescue
18
 
19
  A CenterNet-style, anchor-free person detector for downward-looking drone
20
  cameras: **136,547 parameters, 197 KB full-int8 TFLite**, or **37,363 parameters,
21
  75 KB** for the nano variant, built for ESP32-S3-class microcontrollers. It marks where people are in a 160×160 RGB
22
+ tile so a drone can flag places worth a closer look for a rescue team.
 
23
 
24
+ > **Türkçe özet:** Arama kurtarma dronları için, mikrodenetleyicide
25
+ > çalışacak kadar küçük (136 bin parametre, int8 dosyası 197 KB) bir insan
26
+ > tespit modeli. HERIDAL test setinde int8 hâli, eşik 0,2'de insanların
 
27
  > %83,9'unu buluyor. 75 KB'lık nano sürümü %81,2'sini buluyor. Eğitimde
28
  > kullanılmayan bir WiSARD klibinde bu oranlar %61 ve %62'ye düşüyor. CC BY 4.0.
29
 
 
107
 
108
  ## Data and license
109
 
110
+ **Not trained on the Ferhat Tatık footage.** That footage is licensed to this project for non-commercial use only, so models trained on it are published separately as [VoxHOD-NC](https://huggingface.co/inviolus/VoxHOD-NC) (CC BY-NC 4.0). VoxHOD stays CC BY 4.0.
111
+
112
  Trained on [HERIDAL](https://zenodo.org/records/5662351) (CC BY 3.0):
113
  D. Božić-Štulić, Ž. Marušić, S. Gotovac, "Deep Learning Approach in Aerial
114
  Imagery for Supporting Land Search and Rescue Missions", IJCV 2019.
 
121
 
122
  ## Intended use
123
 
124
+ Search and rescue: helping find people who are lost or injured. It is not
125
+ intended for targeting or harming people.