- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi
So I took the COCO SSD 300 model from here:
https://github.com/weiliu89/caffe/tree/ssd
File here:
https://drive.google.com/open?id=0BzKzrI_SkD1_dUY1Ml9GRTFpUWc
I changed the output from:
layer {
name: "detection_out"
type: "DetectionOutput"
bottom: "mbox_loc"
bottom: "mbox_conf_flatten"
bottom: "mbox_priorbox"
top: "detection_out"
include {
phase: TEST
}
detection_output_param {
num_classes: 81
share_location: true
background_label_id: 0
nms_param {
nms_threshold: 0.45
top_k: 400
}
save_output_param {
label_map_file: "data/coco/labelmap_coco.prototxt"
}
code_type: CENTER_SIZE
keep_top_k: 200
confidence_threshold: 0.01
}
}
To this:
layer {
name: "detection_out"
type: "DetectionOutput"
bottom: "mbox_loc"
bottom: "mbox_conf_flatten"
bottom: "mbox_priorbox"
top: "detection_out"
include {
phase: TEST
}
detection_output_param {
num_classes: 81
share_location: true
background_label_id: 0
nms_param {
nms_threshold: 0.45
top_k: 100
}
code_type: CENTER_SIZE
keep_top_k: 100
confidence_threshold: 0.3
}
}
The accuracy seems low but the main issue is it runs at about 1.5 seconds per frame. Any ideas why?
Is it because there are the full 91 classes?
To get the good performance of the 20 class SSD Demo file will I need to retrain with 20 classes only?
Cheers,
Dan
Link Copied
0 Replies
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page