doc update
doc update
doc update
[REL]Release ready code with objProp/num_chop_heads fix.
[REL]Tested inference as well as training for the release.
placed import of propagate_obj under a condition
docs update
disabling chop_num_heads as it seems to be not working
bugfix for use_image_list in image_segmentation.py
minor license file update
fix parameter override by default
scripts updated to assert for improper CAFFE_ROOT setting and other fixes
fix for training on CPU
added batch_size parameter to several training scripts
fixing the default model name in train_imagenet_classification.sh
mobilenet object detect - 512 intermediate channels, default in the script is now 512x512 size
mobiledetnet object detection model removed as this is not the latest configuration
doc update for MobileNet based object detection
mobilenet object detection script update
mobilenetv2 fix for the first convolution layer - not widening required for first layer
reduced complixity and hw friendly mobiledetnet (initial heads with less channels, 2x2 pool)
[EXP_FILES]Deleted experimental files from caffe-0.16 br.
[PROP_OBJ]ExpV5. ObjProp trying to match with Kalman
[PROP_OBJ]ExpV4. Improving to match with Kalman under some circumstance.
[PROP_OBJ]Improving to match with Kalman under some circumstances
[PROP_OBJ]While comparing with Kalman
[OBJ_PROP]Debugging issue of obj not propagated for VIRB0003.
This is car entering from bot-left corner.
This is car entering from bot-left corner.
[OBJ_PROP]Stand alone app.
[OBJ_PROP]Stand alone app.
Made get_labelname work wihtout labelmap.prototxt
Made get_labelname work wihtout labelmap.prototxt
[HIDE_FILES] Hidign experimental files from caffe-0.16
[OBJ_PROP] Removed depedency of caffe if exteranl dets are read.
[SSD_REL]Removed experimental files
[SSD_REL]Info about non-sqr res.
[SSD_TRAIN]Added non square model for VOC0712.
Update VOC0712_ObjectDetect_README.md
Added info about setting CAFFE_ROOT
[SEGMENTATION]tranining on new dataset.
[OBJ_PROP]Moved obj prop functionality to new file
[SSD_TRAIN]Creating non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Created non-square aspect ratio version for VOC0712 dataset.
Update VOC0712_ObjectDetect_README.md
Added info about setting CAFFE_ROOT
[SEGMENTATION]tranining on new dataset.
[OBJ_PROP]Moved obj prop functionality to new file
[SSD_TRAIN]Creating non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Created non-square aspect ratio version for VOC0712 dataset.
[REL_SSD]removing other than VOC0712 models
[REL_SSD]Added chopped_hd in training script
[SEGMENTATION]Mapillary tranining.
[OBJ_PROP]Moved obj prop functionality to new file
[SSD_TRAIN]Creating non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Created non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Added non square model for VOC0712.
[REL_SSD]Added chopped_hd in training script
[SEGMENTATION]Mapillary tranining.
[OBJ_PROP]Moved obj prop functionality to new file
[SSD_TRAIN]Creating non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Created non-square aspect ratio version for VOC0712 dataset.
[SSD_TRAIN]Added non square model for VOC0712.
MobileNetV1 segmentation changes added
MobileNetV1 segmentation changes added
mobilenetv2 based segmentation being added
mobilenetv2 based segmentation being added
mobilenet/mobiledetnet minor fix in conv layer name (did not affect mobilenet)
adding depthwise conv back into object detection heads
mobilenet-0.5 with fused bn
mobilenet and mobilenetv2 configurations
mobilenet and mobilenetv2 configurations
mobilenet and mobilenetv2 configurations
Some net surgery utilities added.
mobilenet, width multiple of 8, cleanup of scripts
mobilenet cleanup of scripts
Adding MobilenetV2
Adding MobilenetV2
Adding MobilenetV2
Adding MobilenetV2
motion segmentaton - bugfixes
added accuracy measurements for motion segmentation
added motion segmentation inference
motion segmentation scripts
starting to add motion segmentation
starting to add motion segmentation
infer blobs script update
updated installation instructions
added mobilenet-0.5 ssd object detection model
new trained ssd 768x320 model from caffe-0.16
script updates from caffe-0.16
ssd objectdetect (mobilenet-0.5) trained with auto steps and min_dim = -1
object detect - bugfix for weights used from previous stage
docs update
minor training changes
object detection fixes
object detection fixes and improvements
keeping 3x3 pool for PSP for now.
mobilenet detection - tuning
detection - scripts cleanup
mobilenet object detection - tuning and adjustment
license update
mobilenet object detection - complexity reduction
mobilenet object detection being trained
adding mobilenet scripts and models
scripts cleanup after branch off from caffe-0.16
[REL_SSD]Added chopped_hd in training script
[REL_SSD]removing other than VOC0712 models
[SSD_REL]Protecting NMS under param
Update VOC0712_ObjectDetect_README.md
Removed other than VOC0712 models
Removed other than VOC0712 models
Update README.md
Update ssd_detect_video.py
Remove NMS test mode form the rel version
Remove NMS test mode form the rel version
[REL_SSD]Added TI implmentation of NMS and added chopped head functionality
[SSD_REL]Protecting NMS under param
Update README.md
Update VOC0712_ObjectDetect_README.md
[REL_SSD]setting right links
[SSD_REL]added proper links
[SSD_REL] Removed accuracy numbers from the file name
readme - added back the correct readme file
ssd - removed nms and other cleanup
[VOC0712]sparsifying VOC0712 for the release
[OD_REL]Added info for VOC0712 model.
[REL_SSD]updating main readme
[OD_REL]Updating readme for OD
[REL_SSD]Cleaned up for the release.
[REL_SSD]Added models and inferencing scripts
[OD_REL]Updating readme for OD
[OD_REL]Updating readme for OD
[REL_SSD]fixing table issue.
[REL_SSD]fixing table issue.
[REL_SSD]fixing typo
[OD_REL]Added info for VOC0712 model.
[REL_SSD]updating main readme
[OD_REL]Updating readme for OD
[REL_SSD]Cleaned up for the release.
[REL_SSD]Added models and inferencing scripts
[OD_REL]Updating readme for OD
[OD_REL]Updating readme for OD
[REL_SSD]fixing table issue.
[REL_SSD]fixing table issue.
[REL_SSD]fixing typo
Mixed trained for 512x256
[PSPNET]inference script
Update docs/VOC0712_ObjectDetect_README.md
Update README.md
[SSD_REL]Adding file
[TRAIN]Trying to get to 69% accuracy Manu reported